Textbook Chapter 1 Burnside Rings Serge Bouc Ebook All Chapter PDF

Chapter 1 Burnside rings Serge Bouc
Visit to download the full and correct content document:

https://textbookfull.com/product/chapter-1-burnside-rings-serge-bouc/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Algebra 1 Groups Rings Fields and Arithmetic 1st

Edition Ramji Lal
https://textbookfull.com/product/algebra-1-groups-rings-fields-
and-arithmetic-1st-edition-ramji-lal/
Algebras, Rings and Modules, Volume 2: Non-commutative

Algebras and Rings 1st Edition Michiel Hazewinkel
https://textbookfull.com/product/algebras-rings-and-modules-
volume-2-non-commutative-algebras-and-rings-1st-edition-michiel-
hazewinkel/
Unforgettable Promise A Steamy Later in Life Romance

The Next Chapter Book 1 Jeannette Winters
https://textbookfull.com/product/unforgettable-promise-a-steamy-
later-in-life-romance-the-next-chapter-book-1-jeannette-winters/
An introduction to primate conservation First Edition

Serge A. Wich
https://textbookfull.com/product/an-introduction-to-primate-
conservation-first-edition-serge-a-wich/
Endocrine Board Review Reference Edition SCANNED PDF
Serge A Jabbour
https://textbookfull.com/product/endocrine-board-review-
reference-edition-scanned-pdf-serge-a-jabbour/
Foundations of Commutative Rings and Their Modules 1st

Edition Fanggui Wang
https://textbookfull.com/product/foundations-of-commutative-
rings-and-their-modules-1st-edition-fanggui-wang/
Legend of the Five Rings Core Rulebook 5th Edition Max

Brooke
https://textbookfull.com/product/legend-of-the-five-rings-core-
rulebook-5th-edition-max-brooke/
The Wimpy Kid Movie Diary The Next Chapter Kinney
https://textbookfull.com/product/the-wimpy-kid-movie-diary-the-
next-chapter-kinney/
Statistics for international social work and other

behavioral sciences First Edition Serge Lee
https://textbookfull.com/product/statistics-for-international-
social-work-and-other-behavioral-sciences-first-edition-serge-
lee/
CHAPTER 1
Burnside rings
Serge Bouc
Université Paris 7-Denis Diderot, 2 place Jussieu, 75251 Paris Cedex 05. France
1. Introduction 1
2. Basic properties of G-sets 2
3. The ring structure 4
4. Invariants 18
5. The Mackey and Green functor structure 26
6. The Burnside ring as biset-functor 36
Bibliography 47
1. Introduction
1
Let G be a finite group. The Burnside ring B(G) of the group G is one of the
fundamental representation rings of G, namely the ring of permutation representa-
tions.
It is in many ways the universal object to consider when looking at the category
of G-sets. It can be viewed as an analogue of the ring Z of integers for this category.
It can be studied from different points of view. First B(G) is a commutative
ring, and one can look at is prime spectrum and primitive idempotents. This leads
to various induction theorems (Artin, Conlon, Dress): the typical statement here
is that any (virtual) RG-module is a linear combination with suitable coefficients
of modules induced from certain subgroups of G (cyclic, hypoelementary, or Dress
subgroups).
The Burnside ring is the natural framework to study the invariants attached
to structured G-sets (such as G-posets, or more generally simplicial G-sets). Those
invariants are generalizations for the category of G-sets of classical notions, such
as the Möbius function of a poset, or the Steinberg module of a Chevalley group.
They have properties of projectivity, which lead to congruences on the values of
Euler-Poincaré characteristic of some sets of subgroups of G.
The ring B(G) is also functorial with respect to G and subgroups of G, and
this leads to the Mackey functor or Green functor point of view. There are close
connections between the Burnside ring and the Mackey algebra. The Burnside
Mackey functor is a typical example of projective Mackey functor. It is also a
universal object in the category of Green functors. This leads to decomposition of
the category of Mackey functors for G as a directs sum of smaller abelian categories.
Finally B(G) is also functorial with respect to bisets, and this is leads to the
definition of double Burnside rings. Those rings are connected to stable homotopy
1This is a slightly corrected version [date: 16/06/2016] of the original text published in the
“Handbook of Algebra”, vol 2, pp 739-804 (2000) Elsevier
1
2 1. BURNSIDE RINGS
theory via the Segal conjecture, and they provide tools to study the stable splittings
of the classifying spaces of finite groups.
2. Basic properties of G-sets

2.1. Notation. Let G be a finite group. The category of finite G-sets will
be denoted by G-set. The objects of G-set are the finite sets with a left G-
action, morphisms are G-equivariant maps, and composition of morphisms is the
composition of maps.
If H is a subgroup of G, and x is an element of G, the notation x H stands for
x.H.x−1 , and the notation H x for x−1 .H.x. The normalizer of H in G is denoted
by NG (H). If X is a G-set, the stabilizer in G of an element x of X is denoted
by Gx .
The set of conjugacy classes of subgroups of G is denoted by sG , and a set
of representatives of sG is denoted by [sG ]. If H and K are subgroups of G, the
notation H =G K (resp. H ⊆G K) means that there is an element x ∈ G such that
H x = K (resp. H x ⊆ K).
The cardinality of a set S is denoted by |S|.
If p is a prime number, the smallest normal subgroup N of G such that G/N is a
p-group is denoted by Op (G). It is the subgroup of G generated by the p0 -elements,
i.e. the elements of order coprime to p.
More generally, if π is a set of primes, the notation Oπ (G) stands for the
smallest normal subgroup N of G such that G/N is a solvable π-group. The group
G is called π-perfect if Oπ (G) = G. If G itself is a π-group, then the group Oπ (G)
is the limit of the derived series of G. In particular, if π is the set of all primes, then
a group is π-perfect if and only if it is perfect, i.e. equal to its derived subgroup.
The trivial group will be denoted by 1l.
2.2. Operations on G-sets. When X is a G set, and H is a subgroup of G,

one can view X as an H-set by restriction of the action. This H-set is denoted by
ResG H X. If f : X → Y is a morphism of G-sets, let ResH f denote the map f viewed
G
as a morphism of H-sets. This defines a restriction functor ResG H : G-set → H-set.

Now if Z is an H-set, the induced G-set IndG H Z is defined as G ×H X, i.e.
the quotient of the cartesian product G × X by the right action of H given by
(g, x).h = (gh, h−1 x) for g ∈ G, h ∈ H, x ∈ X. The left action of G on G ×H X
is induced by its left action on G × X given by g 0 .(g, x) = (g 0 g, x), for g 0 , g ∈ G
and x ∈ X. If f : Z → T is a morphism of H-sets, then IndG H f is the
morphism of
G-sets from IndG G G
H Z to IndH T defined by (IndH f ) (g, x) = g, f (x) . This defines
an induction functor IndG H : H-set → G-set.
Note that if Z is isomorphic to H/K for some subgroup K of H, then IndG HZ
is isomorphic to G/K.
The set of fixed points of H on X is denoted by X H . It is viewed as a NG (H)/H-
set. If f : X → Y is a morphism of G-sets, then the restriction of f to X H is denoted
by f H . It is a morphism of NG (H)/H-sets from X H to Y H , and this defines a fixed
points functor from G-set to NG (H)/H-set.
When H is a normal subgroup of G, any G/H-set can be viewed as a G-set by
inflation, and this operation defines an inflation functor from G/H-ens to G-ens.
Finally, let H be a subgroup of G, and x be an element of G. If Z is an H-set,
then the group x H acts on Z by h.z = hx z, where h ∈ x H and z ∈ Z, and hx z is
computed in the H-set Z. This gives an x H-set denoted by x Z. If f : Z → T is a
morphism of H-sets, then cx,H (f ) is the map f , viewed as a morphism of x H-sets
from x Z to x T . This defines a conjugation functor cx,H : H-set → x H-set.
Those constructions are connected by various identities. Among them:
2. BASIC PROPERTIES OF G-SETS 3
Proposition 2.2.1. Let G be a finite group, and H and K be subgroups of G.

(1) [Mackey formula] If Z is an H-set, then there is an isomorphism of K-sets
G
K IndH Z '
G
ResG IndK x H
K∩x H ResK x ∩H Z
x∈K\G/H
(2) [Frobenius identity] If X is a G-set and Z is an H-set, then there is an

isomorphism of G-sets

X × IndG H Z ' IndH (ResH X).Z
G G
and in particular for any G-set X, there is an isomorphism of G-sets

X × (G/H) ' IndG G
H ResH X
(3) If Z is an H-set, then there is an isomorphism of NG (H)/H-sets

G N (K)/K x
H Z) '
(IndG K
IndNG
x H (K)/K
( Z)K
x∈NG (K)\G/H
K x ⊆H
Proof. (sketch) (1) Let S be a set of representatives in G of double cosets

K\G/H, and consider the map
G
K IndH Z →
G
ResG IndK x H
x∈S
sending (g, z), with g ∈ G and z ∈ Z to the element (k, hz) of the component
x ∈ S, if g can be written g = kxh, for some k ∈ K and h ∈ H. This is the required
isomorphism of K-sets.
(2) Consider the map

X × IndG H Z → IndH (ResH X) × Z
G G

sending the element x, (g, z) of the left hand side, with x ∈ X, g ∈ G, and
z ∈ Z, to the element g, (g −1 x, z) of the right hand side. This is the required
isomorphism of G-sets. The other isomorphism in assertion (2) is the special case
Z = H/H.
(3) Note that (IndG K G G K
H Z) = (ResNG (K) IndH Z) , and use the Mackey formula.
2.3. Characterization of G-sets.
Lemma 2.3.1. Let G be a finite group.
(1) Any G-set is a disjoint union of transitive ones. If X is a transitive G-set,
and if x ∈ X, then the map
gGx ∈ G/Gx 7→ g.x ∈ X
is an isomorphism of G-sets.
(2) If H and K are subgroups of G, then the map f 7→ f (H) is a one to one
correspondence between the set of G-set homomorphisms from G/H to
G/K and the set of cosets xK ∈ G/K such that H ⊆ x K. In particular,
the G-sets G/H and G/K are isomorphic if and only if H and K are
conjugate in G.
Proof. Both assertions are obvious.
One can characterize a G-set up to isomorphism using the following fundamen-
tal theorem of Burnside ([15] Chapter XII Theorem I):
Theorem 2.3.2. [Burnside] Let G be a finite group, and X and Y be finite
G-sets. Then the following are equivalent:
4 1. BURNSIDE RINGS
(1) The G-sets X and Y are isomorphic.

(2) For any subgroup H of G, the sets X H and Y H have the same cardinality.
Proof. It is clear that (1) implies (2), since any G-set isomorphism X → Y
induces a bijection X H → Y H on the sets of fixed points by any subgroup H of G.
To show the converse, observe that it follows from Lemma 2.3.1 that any finite
G-set X can be written up to isomorphism as
G
X= aK (X) G/K
K∈[sG ]
for some aK (X) ∈ N, where aK (X) G/K denotes the disjoint union of aK (X)
copies of G/K.
Now if (2) holds, for any H ∈ [sG ], there is an equation
X
aK (X) − aK (Y ) |(G/K)H | = 0
K∈[sG ]
The matrix m of this system of equations is given by

m(H, K) = |(G/K)H | = |{x ∈ G/K | H x ⊆ K}|
for K, H ∈ [sG ]. In particular the entry m(H, K) is non-zero if and only if some
conjugate of H is contained in K.
If the set [sG ] is given a total ordering such that H K implies |H| ≤
|K|, then the matrix m is upper triangular, with non-zero diagonal coefficient
m(H, H) = |NG (H) : H|. In particular m is non-singular, and it follows that
aK (X) = aK (Y ), for any K ∈ [sG ], and the G-sets X and Y are isomorphic.
Definition 2.3.3. The above matrix m (or sometimes its transpose) is called
the table of marks of the group G.
3. The ring structure

3.1. Definition. The following definition of the Burnside ring of the group G
appears in an article of Solomon ([37]):
Definition 3.1.1. [Solomon] The Burnside ring B(G) of G is the Grothendieck
group of the category G-set, for relations given by decomposition in disjoint union
of G-sets. The multiplication on B(G) is induced by the direct product of G-sets.
It means that B(G) is the free Z-module with basis the set of equivalence classes
of finite G-sets, quotiented by relations identifying the class of the disjoint union
X t Y of two G-sets X and Y to the sum of the class of X and the class of Y .
The direct product of G-sets is commutative and distributive with respect to
disjoint union, up to canonical isomorphisms. Hence it induces by bilinearity a
commutative ring structure on B(G). The class of a set • of cardinality 1 is a unit
for this ring structure.
Two finite G-sets A and B have the same image in B(G) if and only if there is
a sequence of finite G-sets Xi and Yi , for 1 ≤ i ≤ n, and an isomorphism of G-sets
G
n G
n G
n

At( Xi ) t ( Yi ) ' B t (Xi t Yi )
i=1 i=1 i=1
Taking fixed points of both sides shows that for any subgroup H of G, one has
|AH | = |B H |, and Burnside’s Theorem 2.3.2 now implies that A and B are isomor-
phic as G-sets. In the sequel, the G-set A and its image in B(G) will be identified.
3. THE RING STRUCTURE 5
It follows from Burnside’s Theorem 2.3.2 that any finite G-set X can be written
uniquely up to isomorphism as
G
X' aH (X)G/H
H∈[sG ]
Hence B(G) is a free Z-module, with basis indexed by elements G/H, for H ∈ [sG ].
In this basis, the multiplication law can be recovered by
X
(3.1.2) (G/H).(G/K) = G/(H ∩ x K)
x∈H\G/K
This follows from Proposition 2.2.1, since by Frobenius identities

(G/H) × (G/K) = IndG G G
H ResH IndK K/K
and since by the Mackey formula

G
ResG G
H IndK K/K = H/(H ∩ x K)
x∈H\G/K
Finally, the operations on G-sets defined in section 2.2 all commute with disjoint
unions. Hence they can be extended to the Burnside ring: the elements of B(G)
can be viewed as a formal differences X − Y of two finite G-sets. If F : G-set →
H-set denotes one of the functors of restriction, induction, fixed points, inflation,
or conjugation, then F induces a group homomorphism still denoted by F from
B(G) to B(H), defined by
F (X − Y ) = F (X) − F (Y )
for any finite G-sets X and Y .
Thus for example, if H is a subgroup of G, there is a restriction homomorphism
H : B(G) → B(H)
ResG
This homomorphism is actually a morphism of rings (with unit).
In the special case H = 1l, since B(H) ' Z, this gives an extension of the
cardinality to a map X 7→ |X| = ResG 1l X from B(G) to Z.
Similarly, there is an induction homomorphism
H : B(H) → B(G)
IndG
This morphism is not a ring homomorphism in general.
If H is a subgroup of G, there is a fixed points homomorphism X 7→ X H from
B(G) to B NG (H)/H , which is actually a ring homomorphism. When H is a
normal subgroup of G, there is an inflation homomorphism
G/H : B(G/H) → B(G)

InfG
which is a ring homomorphism.
Finally, if x is an element of G, there is a conjugation homomorphism Z 7→ x Z
from B(H) to B(x H), which is a ring isomorphism.
The following is an obvious extension of Proposition 2.2.1:
Proposition 3.1.3. Let G be a finite group, and H and K be subgroups of G.
(1) [Mackey formula] If Z ∈ B(H), then in B(K)
X
G
ResG
K IndH Z = IndK x H
x∈K\G/H
6 1. BURNSIDE RINGS
(2) [Frobenius identity] If X ∈ B(G) and Z ∈ B(H), then in B(G)

X.IndG G G
H Z = IndH (ResH X).Z
and in particular for any X ∈ B(G)

X.(G/H) = IndG G
H ResH X

(3) If Z ∈ B(H), then in B NG (H)/H
X N (K)/K x
(IndG K
H Z) = IndNG
x H (K)/K
( Z)K
x∈NG (K)\G/H
K x ⊆H
3.2. Fixed points as ring homomorphisms. The ring B(G) is finitely gen-
erated as Z-module, hence it is a noetherian ring. Burnside’s Theorem 2.3.2 can
be interpreted as follows: each subgroup H of G defines a ring homomorphism
φGH : B(G) → Z by φH (X) = |X |. The kernel of φH is a prime ideal, since Z is
G H G
an integral domain, and the intersection of all those kernels for subgroups H of G
is zero. In particular, the ring B(G) is reduced.
Since φG G
H = φK if H and K are conjugate in G, it follows that the product map
Y Y
Φ= H : B(G) →
φG Z
H∈[sG ] H∈[sG ]
is injective. Moreover this map Φ is a map between free Z-modules having the same
rank. Hence the cokernel of Φ is finite.
The matrix m of QΦ with respect to the basis {G/H}H∈[sG ] and to the canonical
basis {uH }H∈[sG ] of H∈[sG ] Z is the table of marks of the group G. Recall from
Definition 2.3.3 that for H, K ∈ [sG ]
m(H, K) = |G/K H | = |{x ∈ G/K | H x ⊆ K}|
The cardinality of the cokernel of Φ is the determinant of m, hence it is equal to
Y
|Coker(Φ)| = |NG (H) : H|
H∈[sG ]
This cokernel has been described by Dress ([19]):

Theorem 3.2.1. [Dress] Let G be a finite group. For H and K in [sG ], set
n(K, H) = |{x ∈ NG (K)/K | <x, K> =G H}|
P Q
Then the element y = H∈[sG ] yH uH of H∈[sG ] Z is in the image of Φ if and only
if for any K ∈ [sG ]
X
n(K, H)yH ≡ 0 (|NG (K)/K|)
H∈[sG ]
Proof. First let X be a finite G-set, and let y = Φ(X). Then with the notation
of the theorem, one has yH = |X H | for all H ∈ [sG ], thus for any K ∈ [sG ]
X X
n(K, H)yH = |X H ||{x ∈ NG (K)/K | <x, K> =G H}|
H∈[sG ] H∈[sG ]
X
= |X <x,K> |
x∈NG (K)/K
X
= |(X K )x |
x∈NG (K)/K
Now for any finite group L acting on a finite set Z one has
X |Lz |
|L| × |L\Z| = |L| = |{(l, z) ∈ L × Z | l.z = z}|
|L|
z∈Z
X
= |Z l |
l∈L
Applying this for L = NG (K)/K and Z = X K gives

X
n(K, H)yH = |NG (K)/K| |NG (K)\X K | ≡ 0 (|NG (K)/K|)
H∈[sG ]
By linearity, this proves the “only if” part of the theorem.

Applying this for X = G/M , for some M ∈ [sG ], shows that there is a matrix
t indexed by [sG ] × [sG ], with entries in Z, such that for K ∈ [sG ]
X
n(K, H)m(H, M ) = |NG (K)/K|t(K, M )
H∈[sG ]
Now the matrix n is upper triangular, and its diagonal coefficients are equal to 1.
Thus the matrix t is upper triangular, and
t(K, K) = m(K, K)/|NG (K)/K| = 1
In particular t is invertible (over Z). P Q
Now suppose that the element y = H∈[sG ] yH uH ∈ H∈[sG ] Z satisfies all the
congruences of the theorem. Since Coker(Φ) is finite, there exist rational numbers
rM , for M ∈ [sG ], such that
X
y= rM Φ(G/M )
M ∈[sG ]
In other words for each H ∈ [sG ]

X
yH = |G/M H |rM
M ∈[sG ]
Thus for each K ∈ [sG ]

X X X
n(K, H)yH = n(K, H)m(H, M )rM
H∈[sG ] H∈[sG ] M ∈[sG ]
X
= |NG (K) : K| t(K, M )rM
M ∈[sG ]
The left hand side is a multiple of |NG (K) : K| by assumption, hence there exist
integers zK such that for all K ∈ [sG ]
X
t(K, M )rM = zK
M ∈[sG ]
Since t is invertible, it follows that rM ∈ Z for all M , and y ∈ Im(Φ).

3.3. Idempotents. The map Φ of the previous section is an injective map
between free Z-modules having the same rank. Hence tensoring with Q gives a
Q-algebra isomorphism
'
Y
Q ⊗Z Φ : Q ⊗Z B(G) → Q
H∈[sG ]
and in particular the algebra Q ⊗Z B(G) is semi-simple. It will be denoted by

QB(G). The component QφG H of QΦ will still be written X 7→ |X |, so in general
H
8 1. BURNSIDE RINGS
|X H | will be a rational number for X ∈ QB(G). More generally, all the notations
defined for B(G) will be extended without change to QB(G). Q
The inverse image by QΦ of the canonical Q-basis of H∈[sG ] Q indexed by H
0
is denoted by eG G G
H . If H is conjugate to H in G, one also sets eH 0 = eH . With this
notation for any subgroups H and K of G, one has

1 if H =G K
|(eH ) | =
G K
0 otherwise
H , for H ∈ [sG ], is the set of primitive idempotents of QB(G).

The set of elements eG
Those idempotents have been computed explicitly by Gluck ([22]), and later inde-
pendently by Yoshida ([49]). This can be done using the following lemma:
(1) Let H be a subgroup of G. Then for any X ∈ QB(G)
H = |X |eH
X.eG H G
Conversely, if Y ∈ QB(G) is such that X.Y = |X H |Y for any X ∈

QB(G), then Y ∈ QeG H.
(2) Let H be a proper subgroup of G. Then ResG G
H eG = 0. Conversely, if
Y ∈ QB(G) is such that ResH Y = 0 for any proper subgroup H of G,
G
then Y ∈ QeGG.
(3) Let H be a subgroup of G. Then
1
eG
H = IndG H
H eH
|NG (H) : H|
H , for H ∈ [sG ], is a Q-basis of QB(G), thus

Proof. (1) The set of elements eG
for any X ∈ QB(G), there are rational numbers rH , for H ∈ [sG ], such that
X
X= rH eG
H
H∈[sG ]
Taking fixed points of both sides by a subgroup K ∈ [sG ] shows that rK = |X K |.

It follows that for all K ∈ [sG ]
K = |X |eK
X.eG K G
Now let Y be an element of QB(G) verifying X.Y = |X H |Y for any X ∈ QB(G).

K .Y = 0 if K 6=G H, thus Y = |Y |eH is a rational multiple
Then in particular eG H G
G
of eH .
(2) Let H be a proper subgroup of G. Then for any subgroup of K of H
H eG ) | = |(eG ) | = 0
|(ResG G K G K
since obviously |(ResGH X) | = |X | for any G-set X, hence for any X in QB(G).
K K
G
It shows that the restriction of eG to any proper subgroup of G is zero.
Conversely, if the restriction of an element Y to any proper subgroup H of G
is zero, then in particular |Y H | = 0 for such a subgroup, and Y = |Y G |eG
G.
H eH . If X is any element of QB(G), then by

(3) Consider next the element IndG H
Frobenius identity

X.IndG H eH = IndH (ResH X).eH = IndH (|X |.eH ) = |X |IndH eH
H G G H G H H H G H
G
It follows that there is a rational number rH such that
(3.3.2) IndG H G G
H eH = rH eH
By the Mackey formula, the restriction of the left hand side to H is equal to
X X
G H
ResG
H IndH eH = IndH x H H
H∩x H ResH x ∩H eH =
x H
eH
x∈H\G/H x∈NG (H)/H
since the restriction ofeHH to H ∩ H is zero if H 6= H. Moreover if y ∈ G, then

x x
y
for any subgroup K of H

H Ky 1 if K = y H
|(y eH
H ) K
| = |(eH ) | =
0 otherwise
y
Thus y eH H
H = ey H , and finally
H IndH eH = |NG (H) : H|eH

G H
ResG H
Equation 3.3.2 gives then

|NG (H) : H|eH G G G
H = rH ResH eH
Taking fixed points of both sides under H gives rHG
= |NG (H) : H|, and
1
(3.3.3) eG
H = IndG H
H eH
|NG (H) : H|
as was to be shown. As a consequence
(3.3.4) ResG G H
H eH = eH

Theorem 3.3.5. [Gluck] Let G be a finite group. If H is a subgroup of G, then
1 X
eG
H = |K| µ(K, H) G/K
|NG (H)|
K⊆H
where µ is the Möbius function of the poset of subgroups of G, and G/K denotes
1 ⊗ G/K ∈ QB(G).
Proof. One can write
X
eH
H = r(K, H) H/K
K⊆H
y
H = ey H for y ∈ G, one can suppose
where r(K, H) is a rational number. Since y eH H
r( K, H) = r(K, H) for y ∈ G. Taking induction from H to G gives

y y
1 X
(3.3.6) eG
H = r(K, H) G/K
|NG (H) : H|
K⊆H
Now the sum of the elements eG H , for H ∈ [sG ], is equal to • = G/G. Summing
over all subgroups H of G instead of H ∈ [sG ] gives
X |NG (H)| 1 X
G/G = r(K, H) G/K
|G| |NG (H) : H|
H⊆G K⊆H
The coefficient of G/K in the right hand side is equal to the coefficient of G/K 0 ,
for any conjugate K 0 of K in G, and it is equal to
X |H|
r(K, H)
|G|
K⊆H⊆G
This must be equal to zero if K 6= G, and equal to 1 if K = G. Setting r0 (K, H) =

|H|
|K| r(K, H) for K ⊆ H, this gives
X
0 |G : K| = 1 if K = G
r (K, H) =
0 otherwise
K⊆H⊆G
10 1. BURNSIDE RINGS
This shows that r0 (K, H) is equal to µ(K, H), where µ is the Möbius function of
the poset of subgroups of G. Thus m(K, H) = |K| |H| µ(K, H), and Equation 3.3.6
gives
1 X
eG
H = |K| µ(K, H) G/K
|NG (H)|
K⊆H
as was to be shown.
The formulae for primitive idempotents in QB(G) lead to a natural question:

when are these idempotents actually in B(G)? More generally, if π is a set of prime
numbers, let Z(π) denote the subring of Q of irreducible fractions with denominator
prime to all the elements of π. In other words Z(π) is the localization of Z with
respect to the set Z − ∪p∈π pZ. One may ask when the idempotents of QB(G) are
actually in Z(π) B(G). The answer is as follows:
Theorem 3.3.7. [Dress] Let G be a finite group, and π be a set of primes. Let
F be a family of subgroups of G, closed by conjugation in G. Let [F] denote the set
F ∩ [sG ]. Then the following conditions are equivalent:
P
(1) The idempotent H∈[F ] eG H lies in Z(π) B(G).
(2) Let H and K be any subgroups of G such that H is a normal subgroup of
K and the quotient K/H is cyclic of prime order p ∈ π. Then H ∈ F if
and only if K ∈ F.
P
Proof. Suppose first that (1) holds, i.e. that the idempotent e = H∈[F ] eG H
lies in Z(π) B(G). This is equivalent to the existence of an integer m coprime to all
the elements of π, such that me ∈ B(G). Now if H and K are subgroups of G such
that H / K and K/H is cyclic of prime order p, then for any element X of B(G),
one has |X H | ≡ |X K | (p). It follows that
m|eH | ≡ m|eK | (p)
But m|eH | is equal to m if H ∈ F, and it is equal to zero otherwise. Thus if p ∈ π,

it follows that assertion (2) holds, because m 6≡ 0 (p).
Conversely, suppose that (2) holds. Let m be an integer such that me ∈ B(G).
By Dress’s Theorem 3.2.1 this is equivalent to require that for each subgroup K
of G
X X
n(K, H)m|eH | = m n(K, H) ≡ 0 (|NG (K) : K|)
H∈[sG ] H∈[F ]
This can also be written as

X
m |{x ∈ NG (K)/K | <x, K> =G H}| ≡ 0 (|NG (K) : K|)
H∈[F ]
i.e.
(3.3.8) m|{x ∈ NG (K)/K | <x, K> ∈ F }| ≡ 0 (|NG (K) : K|)
If (2) holds, then <x, K> ∈ F if and only if <xπ0 , K> ∈ F, where xπ0 denotes
the π 0 -part of x. The set Γ of π 0 -elements y of the group W = NG (K)/K such
that <y, K> ∈ F is invariant under conjugation in W . Let [Γ] denote a set of
representatives of W -conjugacy classes of Γ. Then:

X
|{x ∈ W | xπ0 ∈ Γ}| = |{x ∈ W | xπ0 = γ}|
γ∈Γ
X
= |CW (γ)π |
γ∈Γ
X |W |
= |CW (γ)π |.
|CW (γ)|
γ∈[Γ]
Now by a theorem of Frobenius (see Corollaire 1 of Théorème 23 of [36]), for each

γ ∈ Γ, there exists an integer lγ such that |CW (γ)π | = lγ |CW (γ)|π . Hence
X |W | X |W |π0
|{x ∈ W | xπ0 ∈ Γ}| = lγ = |W |π lγ ≡ 0 (mod. |W |π ).
|CW (γ)|π0 |CW (γ)|π0
γ∈[Γ] γ∈[Γ]
It follows that |W |π0 |{x ∈ W | xπ0 ∈ Γ}| ≡ 0 (mod. |W |), and in particular
Congruence 3.3.8 holds if m = |G|π0 , for any subgroup K of G. Hence me ∈ B(G),
and Assertion (1) holds.
Corollary 3.3.9. Let G be a finite group, and π be a set of primes. If J is a
π-perfect subgroup of G, set
X
fJG = eG
H
H∈[sG ]
O π (H)=G J
Then the set of elements fJG , for π-perfect elements J of [sG ], is the set of primitive
idempotents of Z(π) B(G).
In particular, the set of primitive idempotents of B(G) is in one to one corre-
spondence with the set of conjugacy classes of perfect subgroups of G.
The group G is solvable if and only if G/G is a primitive idempotent of B(G).
Proof. Let J be a π-perfect subgroup of G, and set
F = {H ⊆ G | Oπ (H) =G J}
Then clearly the family F is closed by conjugation in G, and satisfies condition (2)
of Theorem 3.3.7. Thus fJG lies in Z(π) B(G).
To prove that it is primitive, it suffices to show that the family F has no proper
non-empty subfamily satisfying condition (2) of Theorem 3.3.7. But if F 0 is such a
non-empty subfamily of F, and if H ∈ F 0 , then Oπ (H) ∈ F 0 since the composition
factors of H/Oπ (H) are cyclic groups of prime order belonging to π. Thus J ∈ F 0
since F 0 is closed by conjugation. Now if H 0 ∈ F the group Oπ (H 0 ) is in F 0 , and
H 0 ∈ F 0 by the same argument. Thus F 0 = F, and the idempotent fJG is primitive.
The last assertion of the corollary is the case where π is the set of all prime
numbers.
3.4. Prime spectrum. The prime spectrum of B(G) has been determined
by Dress ([20]):
Theorem 3.4.1. [Dress] Let G be a finite group, let p denote a prime number
or zero, and H be a subgroup of G. Let
IH,p (G) = {X ∈ B(G) | |X H | ≡ 0 (p)}
Then:
(1) The set IH,p (G) is a prime ideal of B(G).
(2) If I is a prime ideal of B(G), then there is a subgroup H of G and an
integer p equal to zero or a prime number, such that I = IH,p (G).
(3) If H and K are subgroups of G, and if p and q are prime numbers or zero,
then IH,p (G) ⊆ IK,q (G) if and only if one of the following holds:
(a) One has p = q = 0, and the subgroups H and K are conjugate in G.
In this case moreover IH,p = IK,q .
(b) One has p = 0 and q > 0, and the groups Oq (H) and Oq (K) are
conjugate in G. In this case moreover IH,p 6= IK,q .
(c) One has p = q > 0, and the subgroups Op (H) and Op (K) are conju-
gate in G. In this case moreover IH,p = IK,q .
Proof. (1) Clearly IH,p is the kernel of the ring homomorphism from B(G)
to Z/pZ mapping X to the class of |X H |. Assertion (1) follows, since Z/pZ is an
integral domain.
(2) Conversely, if I is a prime ideal of B(G), then the ring R = B(G)/I is an
integral domain. Let π : B(G) → R be the canonical projection. Since R 6= {0},
there is a subgroup H of G minimal subject to the condition π(G/H) 6= 0. Taking
the image of equation 3.1.2 by π and using the minimality of H gives
X
π(G/H)π(G/K) = π(G/H)
x∈G/K
H x ⊆K
(since moreover HxK = xK if H x ⊆ K). Since R is an integral domain, and since

π(G/H) 6= 0, it follows that
π(G/K) = |{x ∈ G/K | H x ⊆ K}|1R = |(G/K)H |1R
It follows by linearity that π(X) = |X H |1R for any X ∈ B(G). Let p denote the
characteristic of R. Then p is equal to zero or a prime number. Clearly the kernel
of π, equal to I by definition, is also equal to IH,p (G).
(3) Suppose that H and K are conjugate in G. Then for any X ∈ B(G), one has
|X H | = |X K |, and in particular IH,0 (G) = IK,0 (G).
Now if p is a prime number, and if X is a finite G-set, since
p
(H) H/O p (H)
X H = (X O )
and since H/Op (H) is a p-group, it follows that
p
|X H | ≡ |X O (H)
| (p)
for any X ∈ B(G). In particular IH,0 ⊆ IOp (H),p , and IH,p (G) = IK,p (G) if Op (H)
and Op (K) are conjugate in G. Thus IH,p (G) = IOp (H),p (G) for any H ⊆ G.
Conversely, suppose that IH,p (G) ⊆ IK,q (G). There is a surjective ring homo-
morphism from B(G)/IH,p (G) to B(G)/IK,q (G). Since p is the characteristic of
the ring B(G)/IH,p (G), there are three possible cases:
(a) Either p = q = 0. Then the inclusion IH,0 (G) ⊆ IK,0 (G) means that if X ∈
B(G) is such that |X H | = 0, then |X K | = 0. Now X = |NG (K)|eG K is in B(G) by
Theorem 3.3.5, and |X K | = |NG (K)| 6= 0. Thus |X H | 6= 0, hence H is conjugate
to K in G. Clearly in this case IH,0 (G) = IK,0 (G).
(b) The next possible case is p = 0 and q > 0. In this case if X ∈ B(G) is such
that |X H | = 0, then |X K | ≡ 0 (q). Consider the idempotent
X
G
fO q (K) = eG
L
L∈[sG ]
O q (L)=G O q (K)
G
By Corollary 3.3.9, there is an integer m coprime to q such that X = mfO q (K) is
O q (K)
in B(G). Now |X | ≡ |X
K
| ≡ m 6≡ 0 (q). Thus |X | 6= 0, and it follows that
H
Oq (H) =G Oq (K). The inclusion IH,0 (G) ⊆ IK,q (G) is proper since the respective
quotient rings have characteristic 0 and q.
(c) The last case is p = q > 0. Since IH,0 (G) ⊆ IH,p (G), it follows that IH,0 (G) ⊆
IK,p (G). Hence Op (H) =G Op (K) by the discussion of the previous case. And in
this case IH,p (G) = IK,p (G).
3.5. Application to induction theorems.

Notation 3.5.1. Let G be a finite group. Denote by C(G) the set of cyclic
subgroups of G. Denote by RC (G) the ring of complex characters of G, and set
QRC (G) = Q ⊗Z RC (G).
Theorem 3.5.2. [Artin] Let G be a finite group. Then
X
QRC (G) = IndGH QRC (H)
H∈C(G)
In other words, any complex character of G is a linear combination with rational

coefficients of characters induced from cyclic subgroups of G.
Proof. There is a natural homorphism from the Burnside ring B(G) of G to
the ring RC (G), which maps a G-set X to the associated CG-module CX. This
extends to a map of vector spaces QB(G) 7→ QRC (G). Now the value of the
character of the permutation module CX at the element s of G is the trace of s on
CX, which is equal to the number of fixed points |X s | of s on X.
Let H be a subgroup of G. Since |(eG H ) | is equal to 0 if H is not conjugate in
s
G to the subgroup generated by s, it follows that the image of eG H in QRC (G) is

zero unless H is cyclic. And if H is cyclic, the image of eGH is a linear combination
with rational coefficients of permutations characters IndG K 1, for subgroups K of G.
Taking the image in QRC (G) of the decomposition
X
G/G = eG
H
H∈[sG ]
shows that the trivial character is a linear combination with rational coefficients of
such characters IndGK 1, which are induced from cyclic subgroups K of G. Then if
χ is any character of G
χIndG G G
H 1 = IndH (ResH χ)
and the theorem follows.
Notation 3.5.3. Let p be a prime number, and O be a complete local noether-

ian commutative ring with maximal ideal p and residue field k of characteristic p.
If G is a finite group, an OG-lattice is a finitely generated O-free OG-module. Also
denote by O the trivial OG-lattice of O-rank 1.
The Green ring AO (G) is the Grothendieck group of the category of OG- lat-
tices, for relations given by direct sums decompositions.
As an additive group, it is the quotient of the free abelian group with basis the
set of isomorphism classes of OG-lattices, by the subgroup generated by elements
[M ⊕ N ] − [M ] − [N ], for OG-lattices M and N , where [M ] denotes the class
of M . Since Krull-Schmidt theorem holds for OG-lattices, the group AO (G) is free,
with basis the set of indecomposable OG-lattices. The ring structure on AO (G) is
induced by tensor product (over O) of OG-lattices.
There is a natural ring homomorphism πO from B(G) to AO (G), mapping the
(class of the) finite G-set X to the (class of the) permutation lattice OX. It is
natural to look at the kernel IO (G) of this morphism.
Since AO (G) is torsion-free, the sequence

O π
0 → IO (G) → B(G) → AO (G)
is a split exact sequence of abelian groups.
Hence there is an exact sequence
Qπ
0 → QIO (G) → QB(G) →O QAO (G)
It follows that there exists a set NO (G) of subgroups of G, closed by conjugation,
such that X
QIO (G) = QeG
H
H∈NO (G)/G
or equivalently
IO (G) = {X ∈ B(G) | ∀H ⊆ G, H ∈
/ NO (G) ⇒ |X H | = 0}
Clearly the morphism πO commutes with induction and restriction. Now formulae
3.3.3 and 3.3.4 show that eGH ∈ QIO (G) if and only if eH ∈ QIO (H). It follows that
H
there exists a family of finite groups NO such that

NO (G) = {H ⊆ G | H ∈ NO }
The main result of this section is a characterization of the family NO . First a
definition:
Definition 3.5.4. Let p be a prime number. If G is a finite group, let Op (G)
denote the largest normal p-subgroup of G. The group G is called p-hypoelementary,
or cyclic modulo p, if the quotient group G/Op (G) is cyclic. The family of such
finite groups is denoted by Zp . The set of p-hypoelementary subgroups of a finite
group H is denoted by Zp (G).
It turns out that the family NO depends only on p, and is equal to the com-
plement of Zp , by the following theorem (see [17]):
Theorem 3.5.5. [Conlon] Let G be a finite group, and X, Y be finite G-sets.
The following conditions are equivalent:
(1) The OG-lattices OX and OY are isomorphic.
(2) For any p-hypoelementary subgroup H of G, the sets X H and Y H have
the same cardinality.
Proof. Recall that if M is an OG-lattice, then a vertex of M is a subgroup P of
G, minimal such that M is a direct summand of IndG G
P ResP M . The deep argument
for theorem 3.5.5 relies on the Green correspondence, which, for any p-subgroup
P of G, is a bijection between the set of isomorphism classes of indecomposable
OG-lattices with vertex P and the set of isomorphism classes of indecomposable
ONG (P )-lattices with vertex P (see [1] Chapter 3.12).
In the case of permutation modules, or more generally of their direct summands
(or p-permutation modules), there is a more elementary approach, due to Broué
([12]). It relies on the Brauer construction: if M is an OG-module, and P is a
subgroup of G, then the Brauer quotient M [P ] of M at P is defined by
X
M [P ] = M P pM P + P
T rQ (M Q )
Q⊂P
P
where the sum runs over proper subgroups Q of P , and T rQ : M Q → M P is the
relative trace map or transfer map, defined by
X
∀m ∈ M Q , T rQP
(m) = xm
x∈P/Q
If S is a Sylow p-subgroup of P , and if m ∈ M P , then m = T rSP ( |P1:S| m), thus

M [P ] = 0 if P is not a p-subgroup of G. When P is a p-subgroup of G and M = OX
is a permutation lattice associated to a G-set X, the image of the set X P in M [P ]
is a k-basis of M [P ].
In this case moreover, the value of the Brauer character of M at the element
s of G is equal to the number of fixed points of s on X. Hence if s ∈ NG (P ),
the value of the Brauer character of the kNG (P )/P -module M [P ] at the element
sP ∈ NG (P )/P is equal to |X HP,s |, where HP,s = P <s> is the subgroup of G
generated by s and P . Note that HP,s is cyclic modulo p, and that conversely,
if H is a p-hypoelementary subgroup of G, there is a p-subgroup P of G and an
s ∈ NG (P ) such that H = HP,s .
It follows that if OX and OY are isomorphic, then for any p-hypoelementary
subgroup H of G, one has |X H | = |Y H |.
Conversely, if |X H | = |Y H | for any p-hypoelementary subgroup H of G, then
for any p-subgroup P of G, the modules kX P ' OX[P ] and kY P ' OY [P ] have
the same Brauer character. The next proposition shows that OX and OY are
isomorphic, and this completes the proof of Theorem 3.5.5.
Proposition 3.5.6. Let G be a finite group, and let M and N be p-permutation
OG-lattices. Then the following are equivalent:
(1) The OG-lattices M and N are isomorphic.
(2) For any p-subgroup P of G, the kNG (P )/P -modules M [P ] and N [P ] have
the same Brauer character.
Proof. It is clear that (1) implies (2). The proof of the converse is by induction
on the cardinality of the set
S(M ) = {P ⊆ G | M [P ] 6= 0}
Note that if (2) holds, then S(M ) = S(N ).
If S(M ) = ∅, then M [1] = M/pM = N/pN = 0, hence M = N = 0. If
S(M ) 6= ∅, choose a maximal element P of S(M ). Write M = MP ⊕ M 0 (resp.
N = NP ⊕ N 0 ), where all direct summands of MP (resp. NP ) have non-zero
Brauer quotient at P , and where M 0 [P ] = 0 (resp. N 0 [P ] = 0). Theorem 3.5.7
below shows that M [P ] = MP [P ] and N [P ] = NP [P ] are projective kNG (P )/P -
modules. By assumption, they have the same Brauer character, hence they are
isomorphic, and then Theorem 3.5.7 shows that MP and NP are isomorphic. Now
(2) holds for M 0 and N 0 , and moreover |S(M 0 )| < |S(M )|, since S(M 0 ) ⊆ S(M )
and P ∈ S(M )−S(M 0 ). Hence M 0 ' N 0 by induction hypothesis, and M ' N .
Theorem 3.5.7. [Broué] Let G be a finite group, and P be a p-subgroup of G.
(1) Let M be an indecomposable p-permutation OG-module with vertex P .
Then for any subgroup Q of G, the module M [Q] is non-zero if and only
if Q is conjugate to a subgroup of P .
(2) The correspondence M 7→ M [P ] induces a bijection between the set of iso-
morphism classes of indecomposable p-permutation OG-lattices with ver-
tex P and the set of isomorphism classes of indecomposable projective
kNG (P )/P -modules.
Proof. (1) Recall (Higman criterion) that if H is a subgroup of G and if
M is an OG-module, then M is a direct summand of IndG G
H ResH M if and only
P an OH-endomorphism
G
if there exists φ of M such that IdM = T rH (φ), where
−1
G
T rH (φ) = g∈G/H gφg . If S is a Sylow p-subgroup of H, then |H : S| is
invertible in O, and IdM = T rSG (φ/|H : S|). In particular, any vertex of M is a
p-group.
Recall also that if P is a p-group, and Q is a subgroup of P , then the permuta-

tion OP -lattice IndP
Q O is indecomposable, since its reduction modulo p is isomor-
phic to IndQ k, which has simple socle k. Hence any p-permutation OP -lattice is a
P
permutation OP -lattice.
Let M be an indecomposable p-permutation OG-lattice. Then M is a direct
summand of a permutation lattice OX, for some finite G-set X. Let P be a p-
subgroup of G such that M is a direct summand of IndG G G
P ResP M . Then ResP M is
a direct summand of OResP X, hence it is a permutation OP -lattice, and there is
G
a set S of subgroups of P such that

P M ' ⊕ IndQ O
P
ResG
Q∈S
Thus M is a direct summand of some module IndG Q O, for Q ∈ S. Since IndQ O is

P
P M , and since O is a direct summand of ResQ IndQ O, it

P
a direct summand of ResG P
follows that O is a direct summand of ResQ M , and that IndQ O is a direct summand
G G
of IndG G
Q ResQ M .
Now if P is a vertex of M , then Q = P ∈ S. Hence M is a direct summand
P O, and O is a direct summand of ResP M (in other words the module M
of IndG G
has trivial source). It follows that M [Q] 6= 0 for any subgroup of Q. Conversely,
if M [Q] 6= 0 for some subgroup Q of G, then (IndG P O)[Q] 6= 0, thus Q has a fixed
point on the set G/P , i.e. Q ⊆G P . Assertion (1) follows. It shows in particular
that all the vertices of M are conjugate in G.
(2) Since IndG P O ' IndNG (P ) ONG (P )/P , and since the ONG (P )/P is a direct
G
sum of indecomposable projective ONG (P )/P -lattices, it follows from (1) that if
M is an indecomposable p-permutation OG-lattice with vertex P , then there is an
indecomposable projective ONG (P )/P -lattice E such that M is a direct summand
of IndG
NG (P ) E. It is easy to check by Mackey formula that
X
(IndG NG (P ) E)[P ] ' E[P x P/P ] = E/pE
x∈G/NG (P )
P x ⊆NG (P )
since E[Q/P ] = 0 for any non-trivial subgroup Q/P of NG (P )/P .

Now E/pE is an indecomposable projective kNG (P )/P -module, having M [P ]
as a non-zero direct summand. It follows that M [P ] ' E/pE is an indecompos-
able projective kNG (P )/P -module, and the correspondence of assertion (2) is well
defined.
Let E be an indecomposable projective ONG (P )/P -lattice. Write
0
NG (P ) E ' L ⊕ L
IndG
where all the direct summands of L have vertex conjugate to P in G, and no direct
summands of L0 have vertex conjugate to P . Since L0 is a direct summand of IndGP O,
all the indecomposable direct summands of L0 have vertex strictly contained in P
(up to G-conjugation). Hence L0 [P ] = 0, and L[P ] ' E/pE is indecomposable.
It follows that L is indecomposable, and that it is the only indecomposable direct
summand of IndG NG (P ) E with vertex P , up to isomorphism.
Thus if M is an indecomposable p-permutation module with vertex P , and if E
is the only indecomposable projective ONG (P )/P -lattice such that M [P ] ' E/pE,
then M is isomorphic to the only indecomposable direct summand of IndG NG (P ) E
with vertex P . This shows that the correspondence M 7→ M [P ] of assertion (2) is
injective.
Conversely, if E is an indecomposable projective kNG (P )/P -module, then there
is a projective ONG (P )/P -lattice E such that E ' E/pE. In particular E is
indecomposable. Now IndG NG (P ) E has a unique indecomposable direct summand

M with vertex P , and M [P ] ' E. This shows that the correspondence M 7→ M [P ]
of assertion (2) is surjective, and completes the proof of Theorem 3.5.7.
Corollary 3.5.8. Let G be a finite group. Then
X
QAO (G) = H QAO (H)
IndG
H∈Zp (G)
Two OG-lattices M and N are isomorphic if and only if for any p-hypoelementary
subgroup H of G, the restrictions ResG G
H M and ResH N are isomorphic.
Proof. The image by QπO (eG H ) of the idempotent eH of QB(G) in the ring
G
QAO (G) is zero if H is not p-hypoelementary. Thus

X
O = QπO (G/G) = QπO (eGH)
H∈Zp (G)
Now eG
H is a linear combination of elements G/K, for subgroups K of H. Since
subgroups of p-hypoelementary groups are p-hypoelementary, it follows that there
exists rational numbers rK such that
X
O= KO
rK IndG
K∈Zp (G)
since moreover QπO (G/K) is the (class of) the permutation lattice IndG K O. Ten-
soring this identity with M over O, and using Frobenius identity, it follows that
X
M= rK IndG G
K ResK M
K∈Zp (G)
This proves both assertions of the corollary.

Remark 3.5.9. A different proof of Theorem 3.5.5 and Corollary 3.5.8 has been
given by Dress ([21]). It is exposed in Curtis-Reiner ([18] Chapter 11.80D).
Definition 3.5.10. Let p and q be (non-necessarily distinct) prime numbers. A
finite group H is called a (p, q)-Dress group if the group Oq (H) is p-hypoelementary.
A (p, q)-Dress subgroup H of a finite group G is a subgroup of G which is a (p, q)-
Dress group. The set of (p, q)-Dress subgroups of a finite group H is denoted by
Dp,q (G).
Theorem 3.5.11. [Dress] Let G be a finite group. Then
X
AO (G) = IndGH AO (H)
H∈Dp,q (G)
any q
Proof. It follows from the expression of the idempotents of the Burnside ring
B(G): if q is a prime and J is a q-perfect subgroup of G, then the idempotent fJG
of Corollary 3.3.9 is mapped to zero by Z(q) πO if J is not p-hypoelementary.
And if J is p-hypoelementary, the idempotent fJG is a linear combination with
coefficients in Z(q) of elements G/K, where K runs through the (p, q)-Dress sub-
groups of G. This shows that there is an integer mq coprime to q and integers nK
such that X
mq O = KO
nK IndG in AO (G)
K∈Dp,q (G)
Setting X
A0O (G) = IndG
H AO (H)
H∈Dp,q (G)
any q
it follows that the quotient AO (G)/A0O (G) is a torsion group, with finite exponent
coprime to q. Since this holds for any prime q, the groups AO (G) and A0O (G) are
equal.
3.6. Further results and references. The group of units of the Burnside
ring has been studied by Matsuda ([29]), Matsuda-Miyata ([30]), and Yoshida
([50]).
Examples of non-isomorphic groups having isomorphic Burnside rings have
been given by Thévenaz ([40]).
The Burnside ring of a compact Lie group has been defined and studied by tom
Dieck ([44], [45], [46]) and Schwänzl ([34], [35]).
General exposition of the properties of Burnside rings can be found in Benson
([1] Chapter 5.4), Curtis-Reiner ([18] Chapter 11), Karpilovsky ([24] Chapter 15),
tom Dieck ([46]).
4. Invariants
The Burnside ring is an analogue for finite G-sets of the ring Z for finite sets
(and Z is actually isomorphic to the Burnside ring of the trivial group). One can
attach various invariants to structured G-sets, such as G-posets or G-simplicial
complexes.
This section is a self-contained algebraic exposition of the properties of those
invariants. The original definitions and methods of Quillen ([32], [33]) are used
throughout, avoiding however the topological part of this material. Thus for exam-
ple no use will be made of the geometric realization of a poset, and the accent will
be put on acyclic posets rather than contractible ones.
In other words, in order to define and state properties of the invariants attached
to finite G-posets in the Burnside ring, one can forget about the fundamental group
of those posets, and consider only homology groups.
4.1. Homology of posets. Let (X, ≤) be a partially ordered set (poset for
short). As usual, if x, x0 are in X, the notation x < x0 means x ≤ x0 and x 6= x0 .
The notation [x, x0 ]X (resp. [x, x0 [X , ]x, x0 ]X , ]x, x0 [X ) stands for the set of elements
z ∈ X with x ≤ z ≤ x0 (resp. x ≤ z < x0 , x < z ≤ x0 , x < z < x0 ). The notation
[x, .[X (resp. ]x, .[X , ]., x]X , ]., x[X ) stands for the set of elements z ∈ X with x ≤ z
(resp. x < z, z ≤ x, z < x).
If n ∈ N, let Sdn (X) denote the set of chains x0 < . . . < xn of elements of
X of cardinality n + 1. The chain complex C∗ (X, Z) is the complex of Z-modules
defined as follows: for n ∈ N, the module Cn (X, Z) is the free Z-module with basis
Sdn (X). The differential dn : Cn (X, Z) → Cn−1 (X, Z) is given by
X
n
dn (x0 , . . . , xn ) = (−1)i (x0 , . . . , x̂i , . . . , xn )
i=0
where (x0 , . . . , x̂i , . . . , xn ) denotes the chain (x0 , . . . , xn ) − {xi }.

The reduced chain complex C̃∗ (X, Z) is the augmented complex obtained by
setting C−1 (X, Z) = Z, the augmentation map d0 : C0 (X, Z) → C−1 (X, Z) sending
each x0 ∈ X to 1 ∈ Z.
The nth homology group of the complex C∗ (X, Z) (resp. C̃∗ (X, Z)) is denoted
by Hn (X, Z), and called the nth homology group (resp. reduced homology group) of
X. A poset X is acyclic if all its reduced homology groups are equal to zero.
More generally, if K is a ring, the nth homology group of X with coefficients
in K is the nth homology group of the complex K ⊗Z C∗ (X, Z). When K is a field,
one has Hn (X, K) = K ⊗ Hn (X, Z).
4. INVARIANTS 19
The Euler-Poincaré characteristic χ(X) of a finite poset X is defined by

X X
χ(X) = (−1)n rankZ Cn (X, Z) = (−1)n |Sdn (X)|
n≥0 n≥0
Similarly, the reduced Euler-Poincaré characteristic χ̃(X) is defined by

X
χ̃(X) = (−1)n rankZ C̃n (X, Z) = χ(X) − 1
n≥−1
If K is a field, then setting kn = dimK Ker(K ⊗Z dn ) for any n ∈ N

dimK Hn (X, K) = dimK K ⊗Z Hn (X, Z) = kn − dimK Im(K ⊗ dn+1 )
= kn + kn+1 − dimK K ⊗Z Cn+1 (X, Z)
It follows that
X X
χ(X) = (−1)n dimK Hn (X, K) χ̃(X) = (−1)n dimK H̃n (X, K)
n≥0 n≥−1
In particular, if X is acyclic, then χ̃(X) = 0. Similarly, if X and Y are finite

posets, and if there is an homotopy equivalence f∗ from the complex C∗ (X, Z) to
the complex C∗ (Y, Z), then χ(X) = χ(Y ), since f induces a group isomorphism
from Hn (X, Z) to Hn (Y, Z), for any n ∈ N.
If X and Y are posets, a map of posets f : X → Y is a map from X to Y such

that f (x) ≤ f (x0 ) whenever x and x0 are elements of X such that x ≤ x0 . If f is such
a map, there is an induced map of chain complexes C∗ (f, Z) : C∗ (X, Z) → C∗ (Y, Z)
defined for n ∈ N by

f (x0 ), . . . , f (xn ) if f (x0 ) < . . . < f (xn )
Cn (f, Z)(x0 , . . . , xn ) =
0 otherwise
One also defines a reduced map C̃∗ (f, Z) : C̃∗ (X, Z) → C̃∗ (Y, Z) by C̃n (f, Z) =
Cn (f, Z) if n ≤ 0, and C̃−1 = IdZ .
If f and g are maps of posets from X to Y , the notation f ≤ g means that
f (x) ≤ g(x) for any x ∈ X. The maps f and g are said to be comparable if either
f ≤ g or g ≤ f .
Lemma 4.1.1. Let f and g be maps of posets from X to Y . If f and g are
comparable, then the maps of complexes C∗ (f, Z) and C∗ (g, Z) are homotopic, as
well as the maps C̃∗ (f, Z) and C̃∗ (g, Z).
Proof. Suppose for instance that f ≤ g. Consider the map hn : Cn (X, Z) →
Cn+1 (Y, Z) defined for n ∈ N by
X
n

hn (x0 , . . . , xn ) = (−1)i f (x0 ), . . . , f (xi ), g(xi ), . . . , g(xn )
i=0

where the sequence f (x0 ), . . . , f (xi ), g(xi ), . . . , g(xn ) is replaced by 0 if it is not
strictly increasing. It is easy to check that
dn+1 ◦ hn + hn−1 ◦ dn = Cn (g, Z) − Cn (f, Z)
thus the maps Cn (f, Z) and Cn (g, Z) are homotopic. A similar argument can be
used for the augmented complexes, the map h−1 being the zero map.
Corollary 4.1.2. [Quillen]
(1) Let X and Y be posets, and let f : X → Y and g : Y → X be maps of
posets. If g ◦f is comparable to IdX and if f ◦g is comparable to IdY , then
the maps of complexes C̃∗ (f, Z) and C̃∗ (g, Z) are mutual inverse homotopy
equivalences between C̃∗ (X, Z) and C̃∗ (Y, Z).
(2) If the poset X has a biggest element, or a smallest element, the chain
complex C̃∗ (X, Z) is contractible.
Proof. The first assertion is a direct consequence of the previous lemma. For
the second one, denote by m the biggest (or the smallest) element of X, and let •
denote a poset of cardinality one. Apply assertion (1) to the unique map f : X → •
and to the map g : • → X sending the unique element of • to m. The result follows,
since the complex C̃∗ (•, Z) is clearly contractible.
4.2. Invariants attached to finite G-posets. The following definition of
the Lefschetz invariants is due to Thévenaz ([39]):
Definition 4.2.1. Let G be a finite group. A G-poset X is a G-set equipped
with an order relation ≤ compatible to the G-action: if x ≤ x0 are elements of X
and if g ∈ G, then gx ≤ gx0 .
If X and Y are G-posets, a map of G-posets f : X → Y is a map such that
f (gx) = gf (x) if g ∈ G and x ∈ X, and such that f (x) ≤ f (x0 ) in Y , whenever
x ≤ x0 in X. If y ∈ Y , then
f y = {x ∈ X | f (x) ≤ y} fy = {x ∈ X | f (x) ≥ y}
Those sets are sub-Gy -posets of (the restriction of ) X to the stabilizer Gy of y in
G.
If x ≤ y are elements of X, the set ]x, y[X is a Gx,y -poset, where Gx,y is the
stabilizer Gx,y of the pair {x, y}. Similarly, the sets ]x, .[X and ]., x[X are Gx -posets.
If X is a G-poset, then for n ∈ N, the set Sdn (X) is a G-set. When X is finite,
the Lefschetz invariant ΛX of X is the element of B(G) defined by
X
ΛX = (−1)n Sdn (X)
n≥0
The reduced Lefschetz invariant Λ̃X is the element of B(G) defined by

Λ̃X = ΛX − G/G
If x < y are elements of X, the Möbius invariant µX (x, y) is defined as the Lefschetz
invariant of the poset ]x, y[X . It is an element of the Burnside ring B(Gx,y ). By
convention, the Möbius invariant µX (x, x) is equal to Gx /Gx .
It follows from those definitions that |ΛX | is equal to the Euler-Poincaré char-
acteristic of X. One can say more:
(1) If X is a finite G-poset, then for any subgroup H of G
(ΛX )H = ΛX H

in B NG (H)/H . In particular |(ΛX )H | = χ(X H ).
(2) If X and Y are finite G-posets, then ΛX = ΛY in B(G) if and only if
χ(X H ) = χ(Y H ) for any subgroup H of G.
Proof. The first assertion is obvious, since Sdn (X)H = Sdn (X H ) for all
n ∈ N. The second one follows from Burnside’s Theorem 2.3.2.
Definition 4.2.3. A G-poset X is called G-acyclic if the poset X H is acyclic
for any subgroup H of G.
The following is a direct consequence of this definition:
Lemma 4.2.4. Let G be a finite group, and X be a finite G-poset. If X is
G-acyclic, then Λ̃X = 0 in B(G).
4. INVARIANTS 21
Proof. This follows from Burnside’s theorem, since |(Λ̃X )H | = χ̃(X H ) = 0

for any subgroup H of G.
Proposition 4.2.5. Let G be a finite group.
(1) Let X and Y be finite G-posets, and let f : X → Y and g : Y → X be
maps of G-posets. If g ◦ f is comparable to IdX and if f ◦ g is comparable
to IdY , then Λ̃X = Λ̃Y in B(G).
(2) If a G-poset X has a biggest element, or a smallest element, then it is
G-acyclic.
Proof. This is a direct consequence of Corollary 4.1.2: for any subgroup H
of G, the restrictions of f and g to the posets X H and Y H verify the hypotheses
of Corollary 4.1.2. Hence χ̃(X H ) = χ̃(Y H ), and the first assertion follows. For the
second, note that the biggest (resp. smallest) element of X is also a biggest (resp.
smallest) element of X H , for any subgroup H of G.
Example 4.2.6. Let f : X → Y be a maps of finite G-posets. Denote by
X ∗f Y the G-poset defined as follows: the underlying G-set is the disjoint union
X t Y of X and Y . The ordering is defined for z and z 0 in X ∗f Y by

 z, z 0 ∈ X and z ≤ z 0 in X
0 0
z≤z ⇔ z, z ∈ Y and z ≤ z 0 in Y

z ∈ X, z ∈ Y, and f (z) ≤ z 0 in Y
0
Let f denote the injection from Y to X ∗f Y , and g denote the map from X ∗f Y
to Y defined by
f (z) if z ∈ X
g(z) =
z if z ∈ Y
Then f and g are maps of G-posets, such that g ◦ f = IdY and IdX∗f Y ≤ f ◦ g. It
follows that ΛX∗f Y = ΛY .
The consequence is the following relation between Λ̃X and Λ̃Y :

Proposition 4.2.7. Let f : X → Y be a map of finite G-posets. Then in B(G)
X
Λ̃Y = Λ̃X + IndGGy (Λ̃f y Λ̃]y,.[Y )
y∈G\Y
X
Λ̃Y = Λ̃X + IndG
Gy (Λ̃fy Λ̃].,y[Y )
y∈G\Y
In particular, if Λ̃f y = 0 for all y ∈ Y (for instance if f y is Gy -acyclic), then

Λ̃X = Λ̃Y in B(G).
Proof. Let n ∈ N. The set Sdn (X ∗f Y ) is the disjoint union of Sdn (X) and
of the set of sequences z0 < . . . < zn for which zn ∈ Y . Such a sequence has a
smallest element y = zi in Y , hence it can be written as
x0 < . . . < xi−1 < y < y0 < . . . < yn−i−1
where x0 < . . . < xi−1 is in Sdi−1 (f y ) (for i = 0, the convention is that Sd−1 (f y )
is a set of cardinality one), and y0 < . . . < yn−i−1 is in Sdn−i−1 (]y, .[Y ) (for i = n,
the convention is that Sd−1 (]y, .[Y ) has cardinality one).
Keeping track of the action of G, this leads to the following isomorphism of
G-sets
!
G Gn

Sdn (X ∗f Y ) = Sdn (X) t IndGyG
Sdi−1 (f ) × Sdn−i−1 (]y, .[Y )
y
y∈G\Y i=0
Taking alternating sums gives the first equality of the proposition. The second one
follows, by considering the map f : X op → Y op between the opposite posets of X
and Y , since clearly Λ̃X op = Λ̃X for any finite group G and any finite G-poset X.
Corollary 4.2.8. Let X be a finite G-poset.

(1) The reduced Lefschetz invariant of X is equal to
X
Λ̃X = −G/G − IndGGx Λ̃]x,.[X
x∈G\X
(2) If x ≤ y in X, then
X
G Gy,z 0 if x < y
IndGx,y Res µ
Gx,y,z X (z, y) =
x,y,z Gx /Gx if x = y
z∈Gx,y \[x,y]
(3) If f : X → Y is a map of finite G-posets, then

X X
ΛX = − Gy Λf y Λ̃]y,.[Y = −
IndG IndG
Gy Λfy Λ̃].,y[Y
y∈G\Y y∈G\Y
Assertion 2) is the reason for the name of the Möbius invariant.
Proof. Assertion (1) follows from the previous proposition, applied to the
inclusion ∅ → X, since Λ̃∅ = −G/G. Assertion (2) follows from assertion (1),
applied to the Gx,y -poset [x, y[X , which has a smallest element x if x < y.
Assertion (3) follows from assertion (1) and Proposition 4.2.7:
X
Λ̃Y = Λ̃X + Gy (Λf y − Gy /Gy )Λ̃]y,.[Y
IndG
y∈G\Y
X X
= ΛX − G/G + Gy (Λf y Λ̃]y,.[Y ) −
IndG IndG
Gy Λ̃]y,.[Y
y∈G\Y y∈G\Y
X
= ΛX + IndG
Gy (Λf y Λ̃]y,.[Y ) + Λ̃Y
y∈G\Y
The second equality in assertion 3) is similar.

Corollary 4.2.9. Let G be a finite group and X be a finite poset. Denote by
X] (resp. X ] ) the set of elements x ∈ X such that Λ̃].,x[X 6= 0 (resp. Λ̃]x,.[X 6= 0)
in B(Gx ). If Y is a sub-G-poset of X such that X] ⊆ Y ⊆ X (resp. X ] ⊆ Y ⊆ X),
then Λ̃Y = Λ̃X in B(G).
Proof. By induction on the cardinality of X: if X = ∅, then X = X] = X ]
and there is nothing to prove. For the inductive step, consider the inclusion map
i : X] → Y . It is a map of G-posets. Moreover if y ∈ X] , then iy has a biggest
element y, hence it is Gy -acyclic, and Λ̃Gy = 0 in B(Gy ). Now if y ∈ / X] , then
Λ̃].,y[X = 0 in B(Gy ) by definition of X] . Moreover in this case
iy =]., y[X ∩X] = {z ∈ X | z < y, Λ̃].,z[X 6= 0 in B(Gz )}
⊇ {z ∈ X | z < y, Λ̃].,z[X 6= 0 in B(Gz ∩ Gy )} = (]., y[X )]
It follows that there are inclusions of Gy -posets
(]., y[X )] ⊆ iy ⊆]., y[X
Moreover |]., y[X | < |X|. By induction hypothesis, it follows that Λ̃iy = Λ̃].,y[X = 0.
Now the corollary follows from Proposition 4.2.7.
4. INVARIANTS 23
4.3. Steinberg invariants.

Definition 4.3.1. Let G be a finite group, and p be a prime number. The
Steinberg invariant Stp (G) of G at p is the reduced Lefschetz invariant of the poset
sp (G) of non-trivial p-subgroups of G, on which G acts by conjugation.
The reason for this terminology is that if G is a finite simple Chevalley group
in characteristic p, then the (virtual) permutation character associated to Stp (G)
is equal up to a sign to the Steinberg character of G.
Proposition 4.3.2. Let G be a finite group, and p be a prime number. Then
Stp (G) = 0 in B(G) if and only if G has a non-trivial normal p-subgroup.

Proof. If Stp (G) = 0, then in particular χ̃ sp (G)G = 0. Thus sp (G)G is
non-empty, and G has a non-trivial normal p-subgroup.
Conversely, suppose that R 6= 1l is a non-trivial normal p-subgroup of G. Let
f be the map from sp (G) to [R, .[sp (G) defined by f (Q) = Q.R, and let g denote
the inclusion map from [R, .[sp (G) to sp (G). Then f and g are maps of G-posets,
and moreover Id ≤ g ◦ f and f ◦ g = Id. Since [R, .[sp (G) has a smallest element, it
follows from Proposition 4.2.5 that Stp (G) = 0 in B(G).
Remark 4.3.3. Quillen has conjectured that sp (G) is contractible if and only
if Op (G) 6= 1l. The above proof shows actually that sp (G) is G-contractible if and
only if Op (G) 6= 1l (see Thévenaz-Webb [42]).
Proposition 4.3.4. Let G be a finite group, and p be a prime number. Let
ap (G) denote the sub-G-poset of sp (G) consisting of non-trivial elementary abelian
p-subgroups of G, and bp (G) denote the sub-G-poset of sp (G) consisting of non-
trivial p-subgroups P of G such that P = Op NG (P ) . Then
Stp (G) = Λ̃ap (G) = Λ̃bp (G) in B(G)
Proof. Let P be a non-trivial p-subgroup of G. Suppose P is not elemen-
tary abelian, and denote by Φ(P ) the Frattini subgroup of P . Let f :]., P [sp (G) →
[Φ(P ), P [sp (G) be the map defined by f (Q) = Q.Φ(P ), and let g be the inclusion
map from [Φ(P ), P [sp (G) to ]., P [sp (G) . Then f and g are maps of NG (P )-posets.
Moreover f ◦ g = Id, and Id ≤ g ◦ f . Now [Φ(P ), P [sp (G) has a smallest element,

thus Λ̃].,P [sp (G) = 0 in B NG (P ) by Proposition 4.2.5. In other words

sp (G) ] ⊆ ap (G) ⊆ sp (G)
and Corollary 4.2.9 shows that Λ̃sp (G) = Λ̃ap (G) .
The otherequality is similar: let P be a non-trivial p-subgroup of G. Set R =
Op NG (P )/P , and suppose R 6= P . Let f :]P, .[sp (G) → sp NG (P )/P defined by

f (Q) = NQ (P )/P , and let g : sp NG (P )/P →]P, .[sp (G) defined by g(Q/P ) = Q.
Then f ◦ g = Id, and g ◦ f ≤ Id. By Proposition 4.2.5, it follows that
N (P )
Λ̃]P,.[sp (G) = InfNGG (P )/P
Stp NG (P )/P

and this is zero since Op NG (P )/P = R/P 6= 1l. Hence Λ̃]P,.[sp (G) = 0 in

B NG (P )/P . In other words
]
sp (G) ⊆ bp (G) ⊆ sp (G)
and the proposition follows from Corollary 4.2.9.
Notation 4.3.5. For the remainder of this section, the group G will be a finite
group, and F will denote a family of subgroups of G such that:
(1) 1l ∈ F.
(2) F is closed by conjugation.

(3) F is closed by products, i.e. if P and Q are elements of F such that P
normalizes Q, then P.Q ∈ F.
If H is a subgroup of G, then F(H) is the set of subgroups of H which are in F.
It is a family of subgroups of H having the above three properties.
Denote by F the family F with the trivial group removed. Then F and F are
ordered by inclusion of subgroups, and they are G-posets. When H is a subgroup
of G, denote similarly by F(H) the set of non-trivial subgroups of H which are
in F.
Definition 4.3.6. The Steinberg invariant StF (G) of G with respect to F is
the reduced Lefschetz invariant of the G-poset F.
Thus if F = sp (G), then StF (G) = Stp (G).
Let G and F as in 4.3.5. If P ∈ F, then StF (G) = 0

P
Lemma 4.3.7.
in B NG (P )/P .
Proof. Let a denote the inclusion map from [P, .[F to F P , and b denote
the map from F P to [P, .[F defined by b(P ) = F P . Then b ◦ a = Id[P,.[F , and
IdF P ≤ a ◦ b. Thus StF (G)P = Λ̃F P = Λ̃[P,.[F = 0 since [P, .[F has a smallest
element P .
Theorem 4.3.8. [Bouc] Let G and F as in 4.3.5. If X ∈ B(G), set
X NG (P )
StF (G, X) = IndG
NG (P ) µF (1l, P ) InfNG (P )/P X
P
P ∈G\F
Then the map X 7→ StF (G, X) is an idempotent group endomorphism of B(G),

and its image is the set of elements X of B(G) such that X P = 0 in B NG (P )/P
for all P ∈ F.
Proof. Clearly if X P = 0 for P ∈ F, then StF (G, X) = X. Hence the only
thing to check is that if P ∈ F, then StF (G, X)P = 0.
By linearity, one can suppose that X is a G-set, viewed as a G-poset for the
discrete ordering. Let Z denote the subposet of X × F consisting of pairs (x, P )
such that P ⊆ Gx . Let a : Z → X defined by a (x, P ) = x. Then a is a map of
G-posets, and ax is isomorphic to F(Gx ).
Now obviously ΛX = X for any dicrete G-poset X. Moreover all the elements
of X are maximal in X, thus Λ̃]x,.[X = −Gx /Gx for all x ∈ X. It follows from
Proposition 4.2.7 that
X
(4.3.9) ΛX = X = ΛZ − IndG
Gx StF (Gx ) (Gx )
x∈G\X

Now let b : Z → F be the map defined by b (x, P ) = P . Then b is a map of
G-posets, and for Q ∈ F
gQ = {(x, P ) ∈ X × F | Q ⊆ P ⊆ Gx }

The maps c : qQ → X Q and d : X Q → gQ defined by c (x, P ) = x and d(x) =
(x, Q) are maps of NG (Q)-posets, such that d ◦ c ≤ IdgQ and c ◦ d = IdX Q . Thus
by Corollary 4.2.8
X
ΛZ = − IndG
NG (Q) µF (1l, Q)X
Q
Q∈G\F
Hence equation 4.3.9 gives

X X
X=− IndG
NG (Q) µF (1l, Q)X
Q
− IndG
Gx StF (Gx ) (Gx )
Q∈G\F x∈G\X
4. INVARIANTS 25
and finally X
StF (G, X) = − IndG
Gx StF (Gx ) (Gx )
x∈G\X
Now for any x ∈ X and any P ∈ F, if P ⊆ Gx , then StF (Gx ) (Gx )P = 0 by Lemma
4.3.7. Theorem 4.3.8 follows from assertion (3) of Proposition 2.2.1.
Remark 4.3.10. Another proof of Theorem 4.3.8 can be found in [5], where
decompositions of B(G) associated to F are constructed.
Corollary 4.3.11. Let G and F as in 4.3.5, let p be a prime number, and let
O be a complete local noetherian commutative ring with residue field of characteris-
tic p. Suppose that F contains the set sp (G) of non-trivial p-subgroups of G. Then
for any finite G-poset X the image of StF (G, X) in AO (G) is a linear combination
of projective OG-lattices.
Proof. Indeed, in this case, if Y is an element of B(G) such that Y P = 0 for
P ∈ F, then in particular Y P = 0 for any non-trivial p-subgroup P of G. It follows
that |Y H | = 0 P
whenever H is a subgroup of G with Op (H) 6= 1l.
Now Y = H |Y H |eG H in QB(G), and by Theorem 3.5.5, the idempotent eH
G
maps to zero in QAO (G) if H is not p-hypoelementary. If H is hypoelementary

and if Op (H) = 1l, then H is a cyclic p0 -group, and the idempotent eG
H is a linear
combination of elements G/K, for subgroups K of H.
Those remarks show that the image of Y in AO (G) is a linear combination of
0
permutation lattices IndGK O, for cyclic p -subgroups K of G. The corollary follows,
since those lattices are OG-projective.
Remark 4.3.12. If X = G/G, then StF (G, X) = −StF (G). Hence the image
of StF (G) in AO (G) is a linear combination of projective OG-lattices. In fact, it has
been shown by Webb ([47]) that the chain complex C̃∗ (F, Zp ) is the direct sum of
a complex of projective Zp G-lattices, and of a split acyclic augmented subcomplex.
Corollary 4.3.13. Let G and F as in 4.3.5, and suppose moreover that F is
closed by taking subgroups. Let X be a finite G-poset.
(1) If P ∈ F, then there exists an integer mP such that
ResG
P StF (G, X) = mP P/1l
(2) Denote by |G|F the l.c.m of the orders of elements of F. Then
X
χ̃(X) + µF (1l, P ) χ̃(X P ) ≡ 0 (|G|F )
P ∈F
and in particular
χ̃(F) ≡ 0 (|G|F )
Proof. Let Y be an element of B(G) such that Y P = 0 for all P ∈ F.
Consider the restriction of Y to an element Q of F. Then
X
ResGQY = |Y R |eQ
R
R
where R runs through a set of representatives of conjugacy classes of subgroups
of Q. Since F is closed by taking subgroups, the only non-zero term in this sum is
obtained for R = 1l. Hence
Q |Y |
Q Y = |Y |e1l =
ResG Q/1l
|Q|
This shows that |Y | ≡ 0 (|Q|), and ResG Q Y is an integer multiple of Q/1l. The
first assertion follows, for Y = StF (G, X). Taking cardinalities gives the first
congruence. The second one follows from Lemma 4.3.7 in the case X = F.
Remark 4.3.14. When F isthe poset sp (G) of non-trivial p-subgroups of G,

the second congruence is χ̃ sp (G) ≡ 0 (|G|p ), and it is due to Brown ([13]). Similar
congruences have been stated by Thévenaz ([38]) and Brown-Thévenaz ([14]).
The computation of the Steinberg invariants of the symmetric groups, or more
generally of wreath products of a finite group with symmetric groups, can be found
in Bouc ([6]). It requires the use of a ring of formal power series with coefficients
in Burnside rings.
5. The Mackey and Green functor structure

The notion of Mackey functor is a formal generalization of the properties of
induction, restriction, and conjugation exposed in section 2.2. The notion of Green
functor keeps track moreover of the ring structure. Both have several equivalent
definitions, that are quickly recalled hereafter. The Burnside functor is a universal
object in this framework also.
5.1. Mackey functors and subgroups. Let G be a finite group, and R be a
ring. Let R-Mod denote the category of R-modules. The first definition of Mackey
functors is due to Green ([23]):
Definition 5.1.1. [Green] A Mackey functor M for the group G over R (or
with values in R-Mod) consists of the following data:
• For each subgroup H of G, an R-module M (H).
• Whenever H ⊆ K are subgroups of G with H ⊆ K, a map of R-modules
H : M (H) → M (K), called transfer or induction, and a map of R-
tK
modules rH K
: M (K) → M (H) called restriction.
• For each subgroup H of G and each element x ∈ G, a map of R-modules
cx,H : M (H) → M (x H), called conjugation.
Those maps are subject to four types of conditions:
• (Triviality conditions) For any subgroup H of G, and any h ∈ H, the
maps tH H
H , rH , and ch,H are equal to the identity map of M (H).
• (Transitivity conditions) If H ⊆ K ⊆ L are subgroups of G, then tL
K ◦tH =
K
tH and rH ◦ rK = rH . If x, y ∈ G, then cy,x H ◦ cx,H = cyx,H .

L K L L
• (Compatibility conditions) If H ⊆ K are subgroups of G and x ∈ G, then

x x
cx,K ◦ tK
H = tx H ◦ cx,H and cx,H ◦ rH = rx H ◦ cx,K .
K K K
• (Mackey axiom) If H ⊆ K ⊇ L are subgroups of G, then

X
K
rH ◦ tK
L = H∩x L ◦ cx,H x ∩L ◦ rH x ∩L
tH L
x∈H\K/L
If M and N are Mackey functors for G over R, then a morphism of Mackey functors
f : M → N is a collection of morphisms of R-modules fH : M (H) → N (H), which
commute to the maps tK K
H , rH and cx,H .
The category of Mackey functors for G over R is denoted by M ackR (G).
Example 5.1.2. The Burnside Mackey functor B is the Mackey functor with
values in Z-Mod which value at H is the Burnside ring B(H). If H ⊆ K are
K K
subgroups of G, then tK K
H = IndH and rH = ResH . The conjugation maps cx,H are
defined by cx,H (Z) = Z for Z ∈ B(H) and x ∈ G.
x
More generally, if R is a ring, the Burnside functor RB is defined by “tensoring

everything with R”, i.e. setting RB(H) = R ⊗Z B(H), and extending the maps tK H,
K
rH and cx,H in the obvious way.
Another example of Mackey functor is the Green ring functor AO , which value
at H is the Green ring AO (H) of OH-lattices defined in section 3.5. The transfer
is given by induction of lattices, and the operations of restriction and conjugation
5. THE MACKEY AND GREEN FUNCTOR STRUCTURE 27
are the obvious ones. It is clear that the morphism πO actually defines a morphism
of Mackey functors from the Burnside functor to the Green ring functor.
5.2. Mackey functors and G-sets. If M is a Mackey functor for G over R,
if H and K are subgroups of G, and if x ∈ G is such that H x ⊆ K, then there are
maps of R-modules
H x ◦ cx−1 ,H : M (H) → M (K)
ax = tK bx = cx,H x ◦ rH x : M (K) → M (H)
K
Moreover, if k ∈ K, then since M is a Mackey functor

x x x
H xk ◦ c(xk)−1 ,H = ck−1 x−1 ,K ◦ tH = ck−1 ,K ◦ cx−1 ,x K ◦ tH = cx−1 ,x K ◦ tH = ax

axk = tK K K K
Similarly bxk = bx . Hence ax and bx only depend on the class xK. The crucial
observation is that the set of classes xK such that H x ⊆ K is in one to one
correspondence with the set of G-sets homomorphisms from G/H to G/K. This
leads to the second definition of Mackey functors, due to Dress ([21]).
Recall that a bivariant functor M from a category C to a category D is a
pair of functors M = (M∗ , M ∗ ), where M∗ is a functor from C to D and M ∗ is a
functor from C to Dop (or a cofoncteur from C to D), which coincide on objects,
i.e. M∗ (C) = M ∗ (C) for all objects C of C. This common value is simply denoted
by M (C).
Definition 5.2.1. [Dress] A Mackey functor M for the group G with values in
the category R-Mod of R-modules is a bivariant functor M = (M∗ , M ∗ ) from the
category of finite G-sets to R-Mod, with the following two properties:
(1) Let X and Y be any finite G-sets, and let iX (resp. iY ) denote the canon-
ical injection from X (resp. Y ) into X t Y . Then the morphisms

M∗ (iX ), M∗ (iY ) : M (X) ⊕ M (Y ) → M (X t Y )
∗
M (iX )
: M (X t Y ) → M (X) ⊕ M (Y )
M ∗ (iY )
are mutually inverse isomorphisms.
(2) Let
a
X −−−−−→  Y
 
 
b c
y y
Z −−−−−→ T
d
be any cartesian (i.e. pull-back) square of finite G-sets. Then
M∗ (b) ◦ M ∗ (a) = M ∗ (d) ◦ M∗ (c)
A morphism of Mackey functors f : M → N is a natural transformation of bivariant
functors, i.e. a collection of morphisms of R-modules fX : M (X) → N (X), for
finite G-sets X, which commute to the maps M∗ (a) and M ∗ (a) for any morphism
of finite G-sets a : X → Y .
If M is a Mackey functor for the first definition, then M yields a Mackey functor
M̂ for this definition by choosing a set of representatives [G\X], for each finite G-set
X, and then setting
M̂ (X) = ⊕ M (Gx )
x∈[G\X]
Let f : X → Y be a morphism of G-sets. If x ∈ [G\X] and y ∈ [G\Y ] are such that
f (x) ∈ Gy, then there exists g ∈ G such that f (x) = gy. In this case Gx ⊆ g Gy ,
and one can define two maps
G
αy,x = tg Gy x ◦ cg,Gx : M (Gx ) → M (Gy )
G
βx,y = cg−1 ,g Gx ◦ rg Gyx : M (Gy ) → M (Gx )
Those maps depend only on x and y, and do not depend on the chosen element g.
Define moreover αy,x = βx,y = 0 for x ∈ [G\X] and y ∈ [G\Y ] if f (x) ∈ / Gy.
Then the map M̂∗ (f ) is defined by the block matrix (αy,x )y∈[G\Y ],x∈[G\X] , and
the map M̂ ∗ (f ) is defined by the block matrix (βx,y )x∈[G\X],y∈[G\Y ] . One can check
that M̂ is a Mackey functor for the second definition.
Conversely, a Mackey functor M̂ for the second definition yields a Mackey
functor M for the first definition, by setting
M (H) = M̂ (G/H) tK K
H = M̂∗ (πH )
K
rH = M̂ ∗ (πH
K
) cx,H = M̂∗ (γx,H )
where the morphism πH K
is the canonical projection G/H → G/K for H ⊆ K, and
γx,H is the isomorphism yH →7 yx−1x H from G/H to G/x H, for x ∈ G.
Example 5.2.2. One can show (see for instance [9] Chapter 2.4) that the
Mackey functor associated to the Burnside functor B (still denoted by B) can be
described as follows: if X is a finite G-set, let G-set↓X denote the category which
objects are the finite G-sets over X, i.e. the pairs (Y, f ), where Y is a finite G-set
and f : Y → X is a map of G-sets. A morphism in G-set↓X from (Y, f ) to (Z, g)
is a map of G-sets h : Y → Z such that g ◦ h = f . The composition of morphisms
is the composition of maps.
Then B(X) is the Grothendieck group of the category G-set↓X , for relation
given by decomposition in disjoint union. If φ : X → X 0 is a morphism of G-sets,
then the map B∗ (φ) sends (Y, f ) to (Y, φ ◦ f ), and the map M ∗ (φ) sends (Z, g) to
the G-set over X defined by the pull-back square
Y −−−−−→ Z

 
 
 g
y y
X −−−−−→ X 0
φ
5.3. Mackey functors as modules. The third definition of Mackey functors
is due to Thévenaz and Webb ([43]), who defined the following algebra:
Definition 5.3.1. [Thévenaz-Webb] Let G be a finite group, and R be a com-
mutative ring. The Mackey algebra µR (G) is the R-algebra with generators tK K
H , rH ,
and cx,H , for subgroups H ⊆ K of G, and x ∈ G, subject to the following relations:
• If
PH is aH subgroup of G, and if h ∈ H, then tH = rH = ch,H .H Moreover
H H
H⊆G tH = 1µR (G) , and if H 6= K are subgroups of G, then tH tK = 0.

K
• If H ⊆ K ⊆ L are subgroups of G, then tL K ◦ tH = tH and rH ◦ rK = rH .

K L K L L
If x, y ∈ G, then cy, H ◦ cx,H = cyx,H .

x
x
• If H ⊆ K are subgroups of G and x ∈ G, then cx,K ◦ tK H = tx H ◦ cx,H and
K
x
cx,H ◦ rH = rx H ◦ cx,K .
K K
• If H ⊆ K ⊇ L are subgroups of G, then

X
K
rH ◦ tKL = H∩x L ◦ cx,H x ∩L ◦ rH x ∩L
tH L
x∈H\K/L
A Mackey functor for G over R is a just a µR (G)-module, and a morphism of

Mackey functors is a morphism of µR (G)-modules.
If M is a Mackey functor over R for the first definition, set
M̃ = ⊕ M (H)
H⊆G
5. THE MACKEY AND GREEN FUNCTOR STRUCTURE 29
This is endowed with an obvious structure of µR (G)-module: for instance, the

0
action of the generator tK H of µR (G) is zero on the component M (H ) of M̃ , if
H 0 6= H, and it is equal to the map tK
H : M (H) → M (K) on the component M (H).
Conversely, being given a µR (G)-module M̃ , one recovers a Mackey functor for
the first definition by setting M (H) = tHH M̃ . The transfer maps, restriction maps,
and conjugation maps for M are obtained by multiplication by the generators of
µR (G) with the same name: for instance, the relations of the Mackey algebra show
that tKH M (H) ⊆ M (K).
Proposition 5.3.2. [Thévenaz-Webb] Let G be a finite group and R be a com-

mutative ring. Then the algebra µR (G) is a finitely generated free R-module, with
K cx,K x rK x , for subgroups H and L of G, for x ∈ H\G/L,
basis the set of elements tH L
and a subgroup K of H ∩ L up to H ∩ x L-conjugacy.

x
Proof. See [43] Propositions (3.2) and (3.3), or [9] Chapter 4.4.
Remark 5.3.3. It follows in particular that the rank over R of µR (G) does not
depend on R. In other words, the algebra µR (G) is isomorphic to R ⊗Z µZ (G). It is
sometimes convenient to define µR (G) = R⊗Z µZ (G) for any ring R (not necessarily
commutative). The case R = µS (G) for a commutative ring S is of interest (see [9]
Chapter 1.2). Not that if R is not commutative, then µR (G) is not strictly speaking
an R-algebra.
The next corollary states a link between the Burnside ring and the Mackey
algebra:
Corollary 5.3.4. Let F G be a finite group, and R be a commutative ring.
Define a G-set ΩG by ΩG = H⊆G G/H. If H and L are subgroups of G, if x ∈ G
and K is a subgroup of H ∩ x L, denote by πH,x,L the map of G-sets defined by
πH,x,L : yK ∈ G/K 7→ (yH, y x L) ∈ Ω2G
Then the R-linear map defined by
K cx,K x rK x ∈ µR (G) 7→ (G/K, πH,x,L ) ∈ RB(ΩG )
tH L 2
is an isomorphism of R-modules.
Proof. It is easy to check that the images of the basis elements of µR (G) form
an R-basis of RB(Ω2G ).
Remark 5.3.5. Of course, the isomorphism of Corollary 5.3.4 is not an algebra
isomorphism, since µR (G) is not commutative in general. One can show (see [9]
Chapter 4.5.1) that the multiplication law it induces on RB(Ω2G ) is given by
xyz
∗ xyz
(V, W ) ∈ B(ΩG ) 7→ V ◦Y W = B∗
2
B (V × W )
xz xyyz

where xyz
xz is the map (x, y, z) ∈ Ω3G 7→ (x, z) ∈ Ω2G , and xyyzxyz
is the map
(x, y, z) ∈ Ω3G 7→ (x, y, y, z) ∈ Ω4G , and V × W is the product for the Green functor
structure, to be defined in the next section, in Example 5.4.4.
5.4. Green functors. Roughly speaking, a Green functor for the finite group
G over the commutative ring R is a “Mackey functor with a compatible ring struc-
ture”. More precisely, there are two equivalent definitions of Green functors. The
first one is due to Green ([23]):
Definition 5.4.1. A Green functor A for G over R is a Mackey functor for G
over R, such that for any subgroup H of G, the R-module A(H) has a structure of
R-algebra (associative, with unit), with the following properties:
Another random document with
no related content on Scribd:
the Society for six months and who were in the employment of the
Society at the date of commencing the fund, if they were sixteen
years of age or upwards and not over fifty years of age in the case of
males or forty-five years of age in the case of females, should be
members of the fund. The proposed scale of contributions to the
scheme was 5 per cent. of the wages or salary received, and
contributors were to be eligible at sixty years of age to retire on
pension if they so desired. At the age of sixty-five for males and fifty
for females they would be eligible to receive annuities ranging from
25 per cent. of their salaries, after ten years’ payment of
contributions, to 85 per cent. of their salaries after having paid
contributions for fifty-one years. To assist in launching the scheme it
was proposed that the Society should make an initial contribution of
£10,000. It was also proposed that the superannuation fund should
be managed by a committee of seven, which committee should
consist of the chairman and three directors for the time being of the
U.C.B.S. and three representatives of the employees, who must have
at least three years’ service with the Society.
When the scheme was brought forward again at the September
meeting of the Society one of the amendments sent in was from St
George Society, and called for the rejection of the scheme in its
entirety. The motion to reject the scheme was seconded by a
representative of the employees, who referred to the “autocratic”
methods of the directors in adopting this scheme and bringing it
forward without consulting the workers. The result was that the
scheme was disapproved. The scheme which was brought forward by
the S.C.W.S. for the superannuation of their employees suffered the
same fate.
THE INTERNATIONAL CO-OPERATIVE
CONGRESS.
It was as the result of an invitation which came from the Scottish
National Co-operative Conference, held in Kilmarnock in 1910, that
the International Co-operative Congress held in Hamburg in the
autumn of that year decided to come to Glasgow for the 1913
Congress rather than go to Vienna, the claims of which city were
strongly urged by Dr Beno Karpeles, in opposition to the claims of
Glasgow, as set forth by Mr James Deans.
The year 1913 was a big year for Scottish Co-operators, for not only
had they to entertain the International Congress in the autumn, but
Scotland was also the location of the British Congress in the summer,
and pessimists were not wanting who thought that the Co-operators
of Scotland had undertaken a task which was too heavy for them
when they decided to entertain two so great Congresses as the British
and the International in one year. Events proved that the pessimists
were wrong, however. One of the first duties which fell to the lot of
the committee which was appointed to make preparations for the
International Congress was that of finding a suitable house for it, but
they did not require to look far afield. The U.C.B.S. were the owners
of the best hall in the city for the purpose for which it was required.
In St Mungo Hall there was ample accommodation for the Congress
itself, while in the adjoining halls ample space for dining the
delegates could be provided. The Bakery was just across the street,
and cooking and waiting facilities were all that could be desired. So it
was decided that in St Mungo Hall the Congress should be held.
The U.C.B.S. undertook all the work of catering for the delegates at
luncheon each day. They also undertook the provision of the
luncheon which was given by the reception committee on the
Saturday, and they themselves provided the entertainment for one of
the afternoons of Congress. It was universally acknowledged that the
International Congress of 1913 was the best International Co-
operative Congress held, and to this happy result the U.C.B.S.
contributed no small share. The event was one which will not readily
be forgotten by those privileged to take part. Alas, that the
expressions of fraternity so freely uttered then should have been so
soon made of no avail by the outbreak of war.
THE HOLIDAY CAMP.
It is to Mr John Dewar, for many years president of the
Renfrewshire Co-operative Conference Association, that the idea of a
Co-operative holiday camp owes its origin. For many years Mr Dewar
was an enthusiastic Volunteer, and his experiences under canvas
during the annual training periods of his regiment impressed him
with the value of this form of holiday. Associated with him in his
propaganda for a camping association on Co-operative lines was Mr
James Lucas, at that time president of the Glasgow and Suburbs
Conference Association, and latterly, also, Mr John Paton, of the
Renfrewshire Conference council, who had been converted to the
idea as the result of a visit to Douglas, I.O.M., where he had seen the
huge city under canvas which for a number of years housed
thousands of holiday-making Lancashire lads every summer.
By 1910 these gentlemen had been able to get their organisation so
far advanced that they had selected a site on the Ayrshire Coast for
their first camp, and had made arrangements with the farmer who
rented the land. At the last moment, however, the landowner stepped
in and vetoed the whole proceedings. This put an end to doing
anything further with regard to a camp during that year, but the
search for a suitable site continued and, at length, the little farm of
Roseland, situated on Canada Hill, Rothesay, overlooking the Bay,
was secured. The farm was for sale, but the committee in charge of
the arrangements considered that purchase was too bold an initial
step to take, so they leased the farm for six months; securing an
option to purchase at the end of that period if they wished.
Here, in the summer of 1911, the first Scottish Co-operative
holiday camp was established. It was rather a primitive affair, that
first camp. The cooking was done in the little farmhouse, while the
campers had their meals in a large marquee. The U.C.B.S. directors
took a keen interest in the camp from the very beginning. The
catering was done by them, and the catering staff were housed in the
little farmhouse.
Primitive though the arrangements were, they appealed to the
campers, who were unanimous in their praise of the beautiful
situation, the pure air, the perfect catering, and the small outlay for
which they secured a perfect holiday. Thus encouraged, the
committee which had promoted the camp proceeded to organise a
Co-operative society to work it, and in this Co-operative society the
U.C.B.S. took out twenty-five shares. The farm was purchased for
£600, and in September the Baking Society increased the number of
their shares to 100.
In 1912 the camp was much better organised than in 1911, but it
was still far from being what its promoters desired to see it. They
were hampered for lack of funds, however, as the Co-operative
societies were showing caution and a lack of faith in the enterprise,
and were not providing the capital necessary to work it properly as
readily as had been expected. The only fault which the committee
found with the site lay in the fact that in dry summers the water
supply was inadequate. The summer of 1912 also showed them that it
was desirable that something more impervious to rain than a
marquee was desirable for the gatherings of campers and, in order
that these two defects might be put right, they applied to the U.C.B.S.
for a loan of £1,000 on the security of the property. This loan was
granted them, and so good use did they make of the power which it
gave them that, before the time came for opening the camp in 1913,
they had put down a huge storage tank for water, capable of storing
20,000 gallons; and had erected a dining hall large enough to dine
several hundred persons.
The camp was a very great success in the third year. Its popularity
was so great that the committee found it quite impossible to provide
accommodation for all who wished to avail themselves of its facilities
for holiday making, and this has been the case in each succeeding
year, notwithstanding the influence of the war. At the end of the
third season, however, the committee of the association came to the
conclusion that, if the camp was to be made the success they believed
it was capable of becoming, some rearrangement of its management
would require to be made, so they invited the Baking Society to take
it over as a going concern and work it themselves.
They explained to the directors of the Baking Society that they
were not taking this step because they disbelieved in its success, but
solely on the ground that they considered that dual control was not
good for discipline and did not make for good management.
The directors of the Baking Society promised to consider the
matter, and the result was they brought forward to the quarterly
meeting, held in March 1914, a recommendation that the camp
should be taken over, and this recommendation was accepted by the
delegates. Since then the camp has been managed by the U.C.B.S.
In 1914 accommodation was provided for 250 persons, and it is
extremely probable that greatly increased accommodation would
have been provided before now had it not been for the intervention
of the war which, by providing another and much more strenuous
form of camping for the past and prospective frequenters of Canada
Hill, prevented for the time being such further developments.
Doubtless, however, with the return of Europe to sanity, such
developments will take place; until, before many years are past,
almost the whole of the Society’s seven-acre estate will be covered in
the summer and autumn months with the picturesque pyramids of
white canvas.
THE SOCIETY’S PROGRESS.
In all its branches, with the exception of the tearooms, the
progress of the Society during these four years had been remarkable.
In 1913, however, the London Street tearoom was closed, and as soon
as the lease of the Union Halls expired they also were given up. At
the beginning of the period the output had averaged 3,820 sacks per
week, while at its end the average output was 4,648, an increase of
848 sacks per week in four years. The aggregate sales for the year
which ended in July 1914 were £692,600. Truly, the Society had
travelled far from the days when a small two-oven bakehouse
sufficed for all its output. The membership now consisted of 201
societies; which was also a contrast to the eight small struggling
societies which had banded themselves together in the last days of
December 1868 to form the Federation.
The time had now arrived when the Federation was to be put to a
more severe test than ever before since it had attained to years of
maturity. Like a thunderclap the war storm which had been
gathering over Europe during the month of July burst on an
astonished world which had almost come to believe war on such a
gigantic scale impossible, and many were the doubts expressed, even
by sincere wellwishers of the Co-operative movement, as to how it
would weather the storm. The next chapter will tell how one
federation kept the Co-operative flag flying and added to the laurels
it had gained by its devoted and loyal service to the interests of the
common people.
CHAPTER XVIII.
BAKING UNDER WAR CONDITIONS.
THE EARLY DAYS OF WAR—KEEPING DOWN PRICES: THE

DIRECTORS’ VIEW—HALF-PAY TO SOLDIER EMPLOYEES
—ARMY CONTRACTS—HORSES AND MOTORS
COMMANDEERED—PRICE OF BREAD ADVANCED—
DEATH OF BISCUIT FACTORY MANAGER—MORE ARMY
CONTRACTS—EXTENSIONS AND ALTERATIONS—M‘NEIL
STREET BLOCK COMPLETED—BUSINESS ORGANISATION
—TEABREAD AND PASTRY SHOPS—THE PRICE OF BREAD
—INCREASING OUTPUT—BISCUIT AND TEABREAD
TRADE—THE SOCIETY AND ITS WORKERS—DELIVERY
DIFFICULTIES—DEATH OF MR DUNCAN M‘CULLOCH.
On the fourth day of August 1914, Great Britain entered on a new

era. There are not awanting cynics who declare that Great Britain is
hardly ever out of war, but however that may be, there never has
been in the world’s history a war which seemed in its early days to be
fraught with so dire consequences for civilisation as did that into
which the nations of Europe plunged in those early days of August
1914. For the first few days it seemed as if chaos had suddenly
developed. The people seemed to go mad with panic, and rushed to
purchase goods wherever these could be procured, careless of the
prices which they were called on to pay. To the traders, large and
small, it must have seemed like the beginning of a trading
millennium, and they took immediate advantage of the position in
which they found themselves. Prices doubled, trebled, and
quadrupled; yet the eager rush to buy continued. The poor bought
what they could, but they soon reached the end of their meagre
resources. Not so the rich, however. Amongst them, the panic
seemed to be even greater than amongst the poorer sections of the
community, and they provisioned their houses as if for a siege,
careless as to who must want if only they had plenty in store.
EDUCATIONAL DEPUTATION TO ENGLAND

EDUCATIONAL DEPUTATION TO ENGLAND
The bakers of Glasgow were also the victims of the profiteering

craze which had suddenly enmeshed the trading public, and at a
meeting which took place on the afternoon of 3rd August, they
proposed that the price of bread should be raised. Mr James Young,
manager of the U.C.B.S., had been present at this meeting and had
opposed the proposal, with the result that it was not carried out. This
action of his was homologated by his board at a special meeting
which was held the same evening, while at another meeting, held
three days later, the committee decided that they would not raise the
price of bread in the meantime. The Baking Society found itself in a
very favourable position at this time. In M‘Neil Street there were
supplies of flour which would keep the bakery going for nine weeks,
while Clydebank had supplies for six, and Belfast for seven weeks.
They were thus in a position to view the situation with more or less
equanimity. The S.C.W.S. were also in a good position so far as
supplies of flour and wheat were concerned, having some nine or ten
weeks’ supplies either on hand or on board ship at the outbreak of
war, and the rapidity with which the German fleet was bottled up
within its own harbours gave security of passage for the time being to
food-carrying vessels. Nevertheless, the general panic of that first
week of war affected the wheat and flour trade as much as it did
many others, and it was not long before flour was selling at a price
which had never been asked for it in the whole course of the
Federation’s history.
These were conditions under which the two great Co-operative
federations were able to show to advantage, and they were conditions
also which proved the wisdom of the Wholesale Society in
establishing its own purchasing agency in Canada; for while outside
millers were asking as much as 65/ a sack for flour in the first week
of the war, the highest price quoted by the Wholesale Society was
37/, and ordinary flour was sold by them at 1/6 per sack advance on
normal price. Thus the two federations were in a position to see to it
between them that on this side of the Atlantic, at least, no undue
advantage was taken of the circumstances created by the war. At the
meeting of the Baking Society’s board, which took place on 1st
September, the committee placed on record their appreciation of the
manner in which the crisis brought about by the war had been met
by the manager, heads of departments, and employees generally, and
desired the manager to convey this expression of the board’s
appreciation to the heads of departments and employees. At the
quarterly meeting of the Society, the directors had a similar
compliment paid to themselves; Mr Shaw, Cambuslang Society,
moving, and Mr Glasse, S.C.W.S., seconding a motion which received
the unanimous endorsement of the delegates, thanking the directors
for the splendid lead which they had given to the purveyors of bread
in the city and far beyond. Mr Shaw drew attention to the importance
and magnificence of what had been done, and thought that the
nation was entitled to recognise what the Co-operative movement
had done, not only now but at all times, in keeping down prices and
in keeping goods pure.
Perhaps the position in which the Federation was placed by the
war and the view of it taken by the directors can be explained most
clearly in the words of Mr Gerrard himself. Addressing the
September quarterly meeting, he stated that the profits for the half-
year which had ended in July had been a little better than usual, but
[3]
“the future was so indefinite that no one could foresee with any
accuracy what would happen. They had been blamed, and were still
being blamed in certain quarters, for not increasing the price of
bread, but they had resolutely refused, thinking it their duty as a
Federation to keep the price of bread and other commodities as low
as possible. Notwithstanding, he understood the price of biscuits had
been increased in some quarters, and this had brought them some
trade that they should have had before. The way in which prices were
rushed up at the commencement of the war had seemed to the board
most unseemly and unpatriotic, and seemed to show that everyone
was trying to make as much as possible out of the war. It was
unfortunate that the price of bread would require to be increased
before the war was over, but until that step was absolutely necessary
they would not increase it, and then only as far as was necessary.
They had received several contracts for bread, and the manner in
which the manager, heads of departments, and the general body of
the employees acted under the stress of these circumstances was
worthy of all praise.”
3. (The Scottish Co-operator, 25th September 1914.)
HALF-PAY TO SOLDIER EMPLOYEES.
At the outbreak of war, a number of the employees had been
members of the Territorial Forces and had been called up at once,
and the board decided to grant them half-pay. Later, when the call
for more men came, many of the employees joined up, and the
directors decided that those who enlisted should be placed on the
same footing as regarded half-pay as were those who had been called
up at the outbreak of war. At this quarterly meeting the directors
were accused by a representative of the Bonus Investment Society of
not being so generous to their employees who were serving with the
Colours as were some other Co-operative societies, but the chairman
was able to satisfy the delegates that, in those cases where generosity
was most necessary—those of wives with large families—the Baking
Society’s scheme was the one which gave the wife the larger total
income.
This meeting was the one at which the method of giving the
minutes in the form of synopses was first adopted, and one of the
delegates took occasion to compliment the secretary on having
produced synopses of the various minutes which gave the delegates
enough information about the business done to enable them to find
out what was taking place. At this meeting also a grant of £100 was
made to the Belgian Relief Fund, the chairman stating that further
grants could be made if occasion demanded. Other donations made
were:—£500 to the War Relief Fund, £100 to the Lord Mayor of
Belfast’s Fund, and £100 to the fund being raised by the Provost of
Clydebank. It was urged very strongly by several of the delegates that
the administration of the relief fund should be in the hands of the
Government. In the minute of the board meeting of 2nd October, the
fact is noted that during the past quarter the Society’s motors had
covered a distance of 125,015 miles. It had certainly been a busy
quarter.
ARMY CONTRACTS.
Immediately the war broke out, the United Baking Society were
inundated with Army contracts. In the first few days they had to
make arrangements for supplying camps all over the country, in
addition to those for which they had held the contracts. These new
camps included those at Perth, Inverness, Falkirk, and Dunblane,
and all were supplied at current prices. Yet, notwithstanding the fact
that the Baking Society were doing much work for the Army in that
first week of the war, they were not allowed to escape from some of
the other discomforts which come to the owners of horses and
vehicles in a nation at war. On the 6th of August 18 horses, two
motors, and one lorry were commandeered for military purposes.
Other large bakers suffered in the same way, and on behalf of all of
them a telegram was sent to the Officer in Command, pointing out
the importance of the work which was being done by the baking
firms in the distribution of food, and complaining that serious
hardship to the people would ensue if bakers’ horses were removed.
With this telegram the committee of the Baking Society associated
themselves. To enable them to overtake their work, the Society
decided to purchase several motors which had been offered to them,
while a circular-letter, explaining the position, was sent out to the
societies. That the War Office officials were not ungrateful for the
assistance which had been rendered them in the emergency of
mobilisation was shown by the fact that at the board meeting which
was held on 7th August, letters of congratulation on the manner in
which the service of bread to two camps had been carried out were
read to the committee.
Among the minor results due to the war may be mentioned the
cancelling of a number of social meetings, excursions, marriages,
etc., and the closing down of the holiday camp at the end of August.
At the beginning of October, another attempt was made by the
Glasgow Master Bakers’ Association to have the price of bread raised,
but again the directors of the Baking Society blocked the way. At the
same time, however, the committee came to the decision that, in
view of all the circumstances, they would not raise any objection
should another overture be made. At the beginning of October, the
situation was again reviewed and, in view of the fact that by the
beginning of the year the price of flour would be greatly in advance of
that then being used, it was agreed to resist no longer a slight
advance in the price of bread. For three months the Society had been
the means of keeping the price of bread at the rate at which it was
being sold at the outbreak of war. If we assume that their action
influenced only two hundred thousand households, and that the
average consumption of bread in each household was only one 2–lb.
loaf per day, this action of the Baking Society was responsible for
saving to these householders in three months no less a sum than
£21,250, and probably double that sum, for that only supposes an
advance of one farthing per loaf. The first advance in the price of
bread took place on 16th November. In Belfast the price of bread had
been advanced ¼d. per 2–lb. loaf on 19th October.
DEATH OF BISCUIT FACTORY MANAGER.
Mr John Gilmour took charge of the Society’s biscuit factory
shortly after it was commenced, and he continued to manage it until
his death, which took place on 9th October 1914. He was highly
respected by both directors and workmen for his devotion to
business and his tact and kindness in dealing with those under him.
He was succeeded by Mr William Ninian, who had the distinction of
being a Bakery-trained man and who was acting as Mr Gilmour’s
assistant at the time of that gentleman’s death.
At the quarterly meeting, the chairman made fitting and
sympathetic reference to the loss which the Society had sustained
through the death of Mr Gilmour. He also referred to the retiral from
active work of one of the Society’s bread bakers, Mr W. Lees, who
had been in the employment of the Society for forty and a half years.
During the quarter an electrically-driven motor had been at work in
Glasgow for the Society. It was proving very satisfactory and the
board, he said, considered that before long these electrically-driven
cars would replace horses on the streets, although petrol-driven cars
would still be used for the long journeys.
At this quarterly meeting, some of the delegates wished to know
why the board had decided that they would no longer resist an
increase in the price of bread seeing that in their minutes they stated
that their contract for flour was still unbroken. The chairman, in
replying, pointed out that flour at the moment was about 10/ higher
in price than before the war, and the board thought it better that the
price should be raised by a halfpenny now than that it should be put
a penny or three halfpence later, when the stock of cheap flour was
completely exhausted. He also stated that eighty-six of the Society’s
employees were then serving in the Army. At this quarterly meeting,
also, the question of granting relief to what were termed “innocent
enemy aliens,” in other words, British women who had married
subjects of nations at war with Britain, and who through the
operation of war were left destitute in this country, was raised
through an appeal on their behalf issued by the International Co-
operative Alliance executive. A proposal was made to grant £25 to
the fund which was being raised, but this was defeated by a two to
one majority. The cake show was held this year as usual, but, as was
to be expected, the sales were not so great as in some former years.
The price had been advanced on the average one penny per pound.
In their report to the delegates to the December quarterly meeting,
the directors stated that they would regard it as an instruction to
make no increase in the prices of their goods so long as they were
able to work without actual loss.
MORE ARMY CONTRACTS.
In September, an arrangement had been entered into with the
Army Authorities by which the Government supplied the flour and
the Baking Society baked it into bread for the troops. In addition,
several contracts for biscuits for the Army were received. The first
three of these totalled 200 tons. The contracts for the baking of
bread for the Army continued until early in 1916, when an intimation
was received from the Quartermaster of the Scottish Command that
the contract would cease, as the Government were now erecting field
bakeries for themselves. Two months later, however, another
contract was entered into with the Scottish Command on the same
terms as formerly. The contracts carried out by Clydebank and
Belfast Branches have already been referred to.
EXTENSIONS AND ALTERATIONS.
For the whole of the first year of war and almost the whole of the
second year, the building department of the Society was kept busy
with the extension to the M‘Neil Street premises. The principal
reason for this extension was the necessity for securing more space
for the biscuit factory, and as soon as practicable this extension was
carried through. A new travelling oven was installed early in 1915 at a
cost of £175, and at the same time a new biscuit cutting machine was
got at a cost of £300. A gas-fired travelling oven was also installed
before the end of the year and another one, built in accordance with
alterations suggested by the bakery staff, was installed in October
1916. In the spring of 1915, it was decided to cover over the space
between bakehouses Nos. 11 and 12 for the purpose of providing
dressing rooms for the workers, and at the same time it was decided
to put fans in each flat for ventilation purposes.
M‘NEIL STREET BLOCK COMPLETED.
Parts of the new block at the Adelphi Street and South York Street
corner were occupied as they were completed. Particularly, the
ground floor was occupied as an extension to the biscuit factory, but
it was not until June 1916 that the entire building was ready for the
official opening ceremony. This was of a very modest character, for,
as the chairman stated, the directors did not think the circumstances
of the time lent themselves to the celebration of the opening of their
splendid addition by means of a grand function, and in this he
believed the board had interpreted the wishes of the delegates.
The dressing room for the female workers, which was situated on
the fourth flat, was fitted up in an excellent manner. Each worker
had a locker of her own, made of cast steel, of which she alone held
the key. Fourteen wash-hand basins were fitted up and eight spray
baths. There was also an excellently appointed sickroom, where a girl
who became unwell suddenly could rest, and also a room fitted up
with the appropriate scientific appliances, where accidents could be
treated at once.
At the quarterly meeting, which took place on the day on which the
new wing of the building was opened, the chairman stated that the
total cost of the land and the buildings on it had been £11,800, while
the total cost of the bakery buildings, including the addition, had
been £131,000; which had been depreciated to the extent of
£52,000.

Textbook Chapter 1 Burnside Rings Serge Bouc Ebook All Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Textbook Chapter 1 Burnside Rings Serge Bouc Ebook All Chapter PDF

Uploaded by

Copyright:

Available Formats

Chapter 1 Burnside rings Serge Bouc

Visit to download the full and correct content document:

Algebra 1 Groups Rings Fields and Arithmetic 1st

Algebras, Rings and Modules, Volume 2: Non-commutative

Unforgettable Promise A Steamy Later in Life Romance

An introduction to primate conservation First Edition

Foundations of Commutative Rings and Their Modules 1st

Legend of the Five Rings Core Rulebook 5th Edition Max

The Wimpy Kid Movie Diary The Next Chapter Kinney

Statistics for international social work and other

2. Basic properties of G-sets

2.2. Operations on G-sets. When X is a G set, and H is a subgroup of G,

as a morphism of H-sets. This defines a restriction functor ResG H : G-set → H-set.

Proposition 2.2.1. Let G be a finite group, and H and K be subgroups of G.

(2) [Frobenius identity] If X is a G-set and Z is an H-set, then there is an

and in particular for any G-set X, there is an isomorphism of G-sets

(3) If Z is an H-set, then there is an isomorphism of NG (H)/H-sets

Proof. (sketch) (1) Let S be a set of representatives in G of double cosets

(1) The G-sets X and Y are isomorphic.

The matrix m of this system of equations is given by

3. The ring structure

This follows from Proposition 2.2.1, since by Frobenius identities

and since by the Mackey formula

G/H : B(G/H) → B(G)

(2) [Frobenius identity] If X ∈ B(G) and Z ∈ B(H), then in B(G)

and in particular for any X ∈ B(G)

This cokernel has been described by Dress ([19]):

Applying this for L = NG (K)/K and Z = X K gives

By linearity, this proves the “only if” part of the theorem.

In other words for each H ∈ [sG ]

Thus for each K ∈ [sG ]

Since t is invertible, it follows that rM ∈ Z for all M , and y ∈ Im(Φ).

and in particular the algebra Q ⊗Z B(G) is semi-simple. It will be denoted by

H , for H ∈ [sG ], is the set of primitive idempotents of QB(G).

Conversely, if Y ∈ QB(G) is such that X.Y = |X H |Y for any X ∈

H , for H ∈ [sG ], is a Q-basis of QB(G), thus

Taking fixed points of both sides by a subgroup K ∈ [sG ] shows that rK = |X K |.

Now let Y be an element of QB(G) verifying X.Y = |X H |Y for any X ∈ QB(G).

H eH . If X is any element of QB(G), then by

since the restriction ofeHH to H ∩ H is zero if H 6= H. Moreover if y ∈ G, then

H IndH eH = |NG (H) : H|eH

Equation 3.3.2 gives then

r( K, H) = r(K, H) for y ∈ G. Taking induction from H to G gives

This must be equal to zero if K 6= G, and equal to 1 if K = G. Setting r0 (K, H) =

The formulae for primitive idempotents in QB(G) lead to a natural question:

m|eH | ≡ m|eK | (p)

But m|eH | is equal to m if H ∈ F, and it is equal to zero otherwise. Thus if p ∈ π,

This can also be written as

(3.3.8) m|{x ∈ NG (K)/K | <x, K> ∈ F }| ≡ 0 (|NG (K) : K|)

representatives of W -conjugacy classes of Γ. Then:

Now by a theorem of Frobenius (see Corollaire 1 of Théorème 23 of [36]), for each

(since moreover HxK = xK if H x ⊆ K). Since R is an integral domain, and since

3.5. Application to induction theorems.

In other words, any complex character of G is a linear combination with rational

G to the subgroup generated by s, it follows that the image of eG H in QRC (G) is

Notation 3.5.3. Let p be a prime number, and O be a complete local noether-

Since AO (G) is torsion-free, the sequence

there exists a family of finite groups NO such that

If S is a Sylow p-subgroup of P , and if m ∈ M P , then m = T rSP ( |P1:S| m), thus

Recall also that if P is a p-group, and Q is a subgroup of P , then the permuta-

a set S of subgroups of P such that

Thus M is a direct summand of some module IndG Q O, for Q ∈ S. Since IndQ O is

P M , and since O is a direct summand of ResQ IndQ O, it